1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <crypto/hash.h> 28 29 #include <linux/fscrypt.h> 30 #include <linux/fsverity.h> 31 32 struct pagevec; 33 34 #ifdef CONFIG_F2FS_CHECK_FS 35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36 #else 37 #define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42 #endif 43 44 enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_BLKADDR, 64 FAULT_MAX, 65 }; 66 67 #ifdef CONFIG_F2FS_FAULT_INJECTION 68 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 69 70 struct f2fs_fault_info { 71 atomic_t inject_ops; 72 unsigned int inject_rate; 73 unsigned int inject_type; 74 }; 75 76 extern const char *f2fs_fault_name[FAULT_MAX]; 77 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 78 #endif 79 80 /* 81 * For mount options 82 */ 83 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 84 #define F2FS_MOUNT_DISCARD 0x00000004 85 #define F2FS_MOUNT_NOHEAP 0x00000008 86 #define F2FS_MOUNT_XATTR_USER 0x00000010 87 #define F2FS_MOUNT_POSIX_ACL 0x00000020 88 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 89 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 90 #define F2FS_MOUNT_INLINE_DATA 0x00000100 91 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 92 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 93 #define F2FS_MOUNT_NOBARRIER 0x00000800 94 #define F2FS_MOUNT_FASTBOOT 0x00001000 95 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 96 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 97 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 98 #define F2FS_MOUNT_USRQUOTA 0x00080000 99 #define F2FS_MOUNT_GRPQUOTA 0x00100000 100 #define F2FS_MOUNT_PRJQUOTA 0x00200000 101 #define F2FS_MOUNT_QUOTA 0x00400000 102 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 103 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 104 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 105 #define F2FS_MOUNT_NORECOVERY 0x04000000 106 #define F2FS_MOUNT_ATGC 0x08000000 107 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 108 #define F2FS_MOUNT_GC_MERGE 0x20000000 109 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 110 111 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 112 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 113 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 114 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 115 116 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 117 typecheck(unsigned long long, b) && \ 118 ((long long)((a) - (b)) > 0)) 119 120 typedef u32 block_t; /* 121 * should not change u32, since it is the on-disk block 122 * address format, __le32. 123 */ 124 typedef u32 nid_t; 125 126 #define COMPRESS_EXT_NUM 16 127 128 /* 129 * An implementation of an rwsem that is explicitly unfair to readers. This 130 * prevents priority inversion when a low-priority reader acquires the read lock 131 * while sleeping on the write lock but the write lock is needed by 132 * higher-priority clients. 133 */ 134 135 struct f2fs_rwsem { 136 struct rw_semaphore internal_rwsem; 137 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 138 wait_queue_head_t read_waiters; 139 #endif 140 }; 141 142 struct f2fs_mount_info { 143 unsigned int opt; 144 int write_io_size_bits; /* Write IO size bits */ 145 block_t root_reserved_blocks; /* root reserved blocks */ 146 kuid_t s_resuid; /* reserved blocks for uid */ 147 kgid_t s_resgid; /* reserved blocks for gid */ 148 int active_logs; /* # of active logs */ 149 int inline_xattr_size; /* inline xattr size */ 150 #ifdef CONFIG_F2FS_FAULT_INJECTION 151 struct f2fs_fault_info fault_info; /* For fault injection */ 152 #endif 153 #ifdef CONFIG_QUOTA 154 /* Names of quota files with journalled quota */ 155 char *s_qf_names[MAXQUOTAS]; 156 int s_jquota_fmt; /* Format of quota to use */ 157 #endif 158 /* For which write hints are passed down to block layer */ 159 int alloc_mode; /* segment allocation policy */ 160 int fsync_mode; /* fsync policy */ 161 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 162 int bggc_mode; /* bggc mode: off, on or sync */ 163 int memory_mode; /* memory mode */ 164 int discard_unit; /* 165 * discard command's offset/size should 166 * be aligned to this unit: block, 167 * segment or section 168 */ 169 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 170 block_t unusable_cap_perc; /* percentage for cap */ 171 block_t unusable_cap; /* Amount of space allowed to be 172 * unusable when disabling checkpoint 173 */ 174 175 /* For compression */ 176 unsigned char compress_algorithm; /* algorithm type */ 177 unsigned char compress_log_size; /* cluster log size */ 178 unsigned char compress_level; /* compress level */ 179 bool compress_chksum; /* compressed data chksum */ 180 unsigned char compress_ext_cnt; /* extension count */ 181 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 182 int compress_mode; /* compression mode */ 183 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 184 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 185 }; 186 187 #define F2FS_FEATURE_ENCRYPT 0x0001 188 #define F2FS_FEATURE_BLKZONED 0x0002 189 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 190 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 191 #define F2FS_FEATURE_PRJQUOTA 0x0010 192 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 193 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 194 #define F2FS_FEATURE_QUOTA_INO 0x0080 195 #define F2FS_FEATURE_INODE_CRTIME 0x0100 196 #define F2FS_FEATURE_LOST_FOUND 0x0200 197 #define F2FS_FEATURE_VERITY 0x0400 198 #define F2FS_FEATURE_SB_CHKSUM 0x0800 199 #define F2FS_FEATURE_CASEFOLD 0x1000 200 #define F2FS_FEATURE_COMPRESSION 0x2000 201 #define F2FS_FEATURE_RO 0x4000 202 203 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 204 ((raw_super->feature & cpu_to_le32(mask)) != 0) 205 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 206 #define F2FS_SET_FEATURE(sbi, mask) \ 207 (sbi->raw_super->feature |= cpu_to_le32(mask)) 208 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 209 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 210 211 /* 212 * Default values for user and/or group using reserved blocks 213 */ 214 #define F2FS_DEF_RESUID 0 215 #define F2FS_DEF_RESGID 0 216 217 /* 218 * For checkpoint manager 219 */ 220 enum { 221 NAT_BITMAP, 222 SIT_BITMAP 223 }; 224 225 #define CP_UMOUNT 0x00000001 226 #define CP_FASTBOOT 0x00000002 227 #define CP_SYNC 0x00000004 228 #define CP_RECOVERY 0x00000008 229 #define CP_DISCARD 0x00000010 230 #define CP_TRIMMED 0x00000020 231 #define CP_PAUSE 0x00000040 232 #define CP_RESIZE 0x00000080 233 234 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 235 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 236 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 237 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 238 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 239 #define DEF_CP_INTERVAL 60 /* 60 secs */ 240 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 241 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 242 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 243 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 244 245 struct cp_control { 246 int reason; 247 __u64 trim_start; 248 __u64 trim_end; 249 __u64 trim_minlen; 250 }; 251 252 /* 253 * indicate meta/data type 254 */ 255 enum { 256 META_CP, 257 META_NAT, 258 META_SIT, 259 META_SSA, 260 META_MAX, 261 META_POR, 262 DATA_GENERIC, /* check range only */ 263 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 264 DATA_GENERIC_ENHANCE_READ, /* 265 * strong check on range and segment 266 * bitmap but no warning due to race 267 * condition of read on truncated area 268 * by extent_cache 269 */ 270 DATA_GENERIC_ENHANCE_UPDATE, /* 271 * strong check on range and segment 272 * bitmap for update case 273 */ 274 META_GENERIC, 275 }; 276 277 /* for the list of ino */ 278 enum { 279 ORPHAN_INO, /* for orphan ino list */ 280 APPEND_INO, /* for append ino list */ 281 UPDATE_INO, /* for update ino list */ 282 TRANS_DIR_INO, /* for transactions dir ino list */ 283 FLUSH_INO, /* for multiple device flushing */ 284 MAX_INO_ENTRY, /* max. list */ 285 }; 286 287 struct ino_entry { 288 struct list_head list; /* list head */ 289 nid_t ino; /* inode number */ 290 unsigned int dirty_device; /* dirty device bitmap */ 291 }; 292 293 /* for the list of inodes to be GCed */ 294 struct inode_entry { 295 struct list_head list; /* list head */ 296 struct inode *inode; /* vfs inode pointer */ 297 }; 298 299 struct fsync_node_entry { 300 struct list_head list; /* list head */ 301 struct page *page; /* warm node page pointer */ 302 unsigned int seq_id; /* sequence id */ 303 }; 304 305 struct ckpt_req { 306 struct completion wait; /* completion for checkpoint done */ 307 struct llist_node llnode; /* llist_node to be linked in wait queue */ 308 int ret; /* return code of checkpoint */ 309 ktime_t queue_time; /* request queued time */ 310 }; 311 312 struct ckpt_req_control { 313 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 314 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 315 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 316 atomic_t issued_ckpt; /* # of actually issued ckpts */ 317 atomic_t total_ckpt; /* # of total ckpts */ 318 atomic_t queued_ckpt; /* # of queued ckpts */ 319 struct llist_head issue_list; /* list for command issue */ 320 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 321 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 322 unsigned int peak_time; /* peak wait time in msec until now */ 323 }; 324 325 /* for the bitmap indicate blocks to be discarded */ 326 struct discard_entry { 327 struct list_head list; /* list head */ 328 block_t start_blkaddr; /* start blockaddr of current segment */ 329 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 330 }; 331 332 /* default discard granularity of inner discard thread, unit: block count */ 333 #define DEFAULT_DISCARD_GRANULARITY 16 334 335 /* max discard pend list number */ 336 #define MAX_PLIST_NUM 512 337 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 338 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 339 340 enum { 341 D_PREP, /* initial */ 342 D_PARTIAL, /* partially submitted */ 343 D_SUBMIT, /* all submitted */ 344 D_DONE, /* finished */ 345 }; 346 347 struct discard_info { 348 block_t lstart; /* logical start address */ 349 block_t len; /* length */ 350 block_t start; /* actual start address in dev */ 351 }; 352 353 struct discard_cmd { 354 struct rb_node rb_node; /* rb node located in rb-tree */ 355 union { 356 struct { 357 block_t lstart; /* logical start address */ 358 block_t len; /* length */ 359 block_t start; /* actual start address in dev */ 360 }; 361 struct discard_info di; /* discard info */ 362 363 }; 364 struct list_head list; /* command list */ 365 struct completion wait; /* compleation */ 366 struct block_device *bdev; /* bdev */ 367 unsigned short ref; /* reference count */ 368 unsigned char state; /* state */ 369 unsigned char queued; /* queued discard */ 370 int error; /* bio error */ 371 spinlock_t lock; /* for state/bio_ref updating */ 372 unsigned short bio_ref; /* bio reference count */ 373 }; 374 375 enum { 376 DPOLICY_BG, 377 DPOLICY_FORCE, 378 DPOLICY_FSTRIM, 379 DPOLICY_UMOUNT, 380 MAX_DPOLICY, 381 }; 382 383 struct discard_policy { 384 int type; /* type of discard */ 385 unsigned int min_interval; /* used for candidates exist */ 386 unsigned int mid_interval; /* used for device busy */ 387 unsigned int max_interval; /* used for candidates not exist */ 388 unsigned int max_requests; /* # of discards issued per round */ 389 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 390 bool io_aware; /* issue discard in idle time */ 391 bool sync; /* submit discard with REQ_SYNC flag */ 392 bool ordered; /* issue discard by lba order */ 393 bool timeout; /* discard timeout for put_super */ 394 unsigned int granularity; /* discard granularity */ 395 }; 396 397 struct discard_cmd_control { 398 struct task_struct *f2fs_issue_discard; /* discard thread */ 399 struct list_head entry_list; /* 4KB discard entry list */ 400 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 401 struct list_head wait_list; /* store on-flushing entries */ 402 struct list_head fstrim_list; /* in-flight discard from fstrim */ 403 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 404 unsigned int discard_wake; /* to wake up discard thread */ 405 struct mutex cmd_lock; 406 unsigned int nr_discards; /* # of discards in the list */ 407 unsigned int max_discards; /* max. discards to be issued */ 408 unsigned int max_discard_request; /* max. discard request per round */ 409 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 410 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 411 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 412 unsigned int discard_granularity; /* discard granularity */ 413 unsigned int undiscard_blks; /* # of undiscard blocks */ 414 unsigned int next_pos; /* next discard position */ 415 atomic_t issued_discard; /* # of issued discard */ 416 atomic_t queued_discard; /* # of queued discard */ 417 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 418 struct rb_root_cached root; /* root of discard rb-tree */ 419 bool rbtree_check; /* config for consistence check */ 420 }; 421 422 /* for the list of fsync inodes, used only during recovery */ 423 struct fsync_inode_entry { 424 struct list_head list; /* list head */ 425 struct inode *inode; /* vfs inode pointer */ 426 block_t blkaddr; /* block address locating the last fsync */ 427 block_t last_dentry; /* block address locating the last dentry */ 428 }; 429 430 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 431 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 432 433 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 434 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 435 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 436 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 437 438 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 439 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 440 441 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 442 { 443 int before = nats_in_cursum(journal); 444 445 journal->n_nats = cpu_to_le16(before + i); 446 return before; 447 } 448 449 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 450 { 451 int before = sits_in_cursum(journal); 452 453 journal->n_sits = cpu_to_le16(before + i); 454 return before; 455 } 456 457 static inline bool __has_cursum_space(struct f2fs_journal *journal, 458 int size, int type) 459 { 460 if (type == NAT_JOURNAL) 461 return size <= MAX_NAT_JENTRIES(journal); 462 return size <= MAX_SIT_JENTRIES(journal); 463 } 464 465 /* for inline stuff */ 466 #define DEF_INLINE_RESERVED_SIZE 1 467 static inline int get_extra_isize(struct inode *inode); 468 static inline int get_inline_xattr_addrs(struct inode *inode); 469 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 470 (CUR_ADDRS_PER_INODE(inode) - \ 471 get_inline_xattr_addrs(inode) - \ 472 DEF_INLINE_RESERVED_SIZE)) 473 474 /* for inline dir */ 475 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 476 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 477 BITS_PER_BYTE + 1)) 478 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 479 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 480 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 481 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 482 NR_INLINE_DENTRY(inode) + \ 483 INLINE_DENTRY_BITMAP_SIZE(inode))) 484 485 /* 486 * For INODE and NODE manager 487 */ 488 /* for directory operations */ 489 490 struct f2fs_filename { 491 /* 492 * The filename the user specified. This is NULL for some 493 * filesystem-internal operations, e.g. converting an inline directory 494 * to a non-inline one, or roll-forward recovering an encrypted dentry. 495 */ 496 const struct qstr *usr_fname; 497 498 /* 499 * The on-disk filename. For encrypted directories, this is encrypted. 500 * This may be NULL for lookups in an encrypted dir without the key. 501 */ 502 struct fscrypt_str disk_name; 503 504 /* The dirhash of this filename */ 505 f2fs_hash_t hash; 506 507 #ifdef CONFIG_FS_ENCRYPTION 508 /* 509 * For lookups in encrypted directories: either the buffer backing 510 * disk_name, or a buffer that holds the decoded no-key name. 511 */ 512 struct fscrypt_str crypto_buf; 513 #endif 514 #if IS_ENABLED(CONFIG_UNICODE) 515 /* 516 * For casefolded directories: the casefolded name, but it's left NULL 517 * if the original name is not valid Unicode, if the original name is 518 * "." or "..", if the directory is both casefolded and encrypted and 519 * its encryption key is unavailable, or if the filesystem is doing an 520 * internal operation where usr_fname is also NULL. In all these cases 521 * we fall back to treating the name as an opaque byte sequence. 522 */ 523 struct fscrypt_str cf_name; 524 #endif 525 }; 526 527 struct f2fs_dentry_ptr { 528 struct inode *inode; 529 void *bitmap; 530 struct f2fs_dir_entry *dentry; 531 __u8 (*filename)[F2FS_SLOT_LEN]; 532 int max; 533 int nr_bitmap; 534 }; 535 536 static inline void make_dentry_ptr_block(struct inode *inode, 537 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 538 { 539 d->inode = inode; 540 d->max = NR_DENTRY_IN_BLOCK; 541 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 542 d->bitmap = t->dentry_bitmap; 543 d->dentry = t->dentry; 544 d->filename = t->filename; 545 } 546 547 static inline void make_dentry_ptr_inline(struct inode *inode, 548 struct f2fs_dentry_ptr *d, void *t) 549 { 550 int entry_cnt = NR_INLINE_DENTRY(inode); 551 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 552 int reserved_size = INLINE_RESERVED_SIZE(inode); 553 554 d->inode = inode; 555 d->max = entry_cnt; 556 d->nr_bitmap = bitmap_size; 557 d->bitmap = t; 558 d->dentry = t + bitmap_size + reserved_size; 559 d->filename = t + bitmap_size + reserved_size + 560 SIZE_OF_DIR_ENTRY * entry_cnt; 561 } 562 563 /* 564 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 565 * as its node offset to distinguish from index node blocks. 566 * But some bits are used to mark the node block. 567 */ 568 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 569 >> OFFSET_BIT_SHIFT) 570 enum { 571 ALLOC_NODE, /* allocate a new node page if needed */ 572 LOOKUP_NODE, /* look up a node without readahead */ 573 LOOKUP_NODE_RA, /* 574 * look up a node with readahead called 575 * by get_data_block. 576 */ 577 }; 578 579 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 580 581 /* congestion wait timeout value, default: 20ms */ 582 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 583 584 /* maximum retry quota flush count */ 585 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 586 587 /* maximum retry of EIO'ed page */ 588 #define MAX_RETRY_PAGE_EIO 100 589 590 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 591 592 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 593 594 /* dirty segments threshold for triggering CP */ 595 #define DEFAULT_DIRTY_THRESHOLD 4 596 597 /* for in-memory extent cache entry */ 598 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 599 600 /* number of extent info in extent cache we try to shrink */ 601 #define EXTENT_CACHE_SHRINK_NUMBER 128 602 603 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 604 #define RECOVERY_MIN_RA_BLOCKS 1 605 606 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 607 608 struct rb_entry { 609 struct rb_node rb_node; /* rb node located in rb-tree */ 610 union { 611 struct { 612 unsigned int ofs; /* start offset of the entry */ 613 unsigned int len; /* length of the entry */ 614 }; 615 unsigned long long key; /* 64-bits key */ 616 } __packed; 617 }; 618 619 struct extent_info { 620 unsigned int fofs; /* start offset in a file */ 621 unsigned int len; /* length of the extent */ 622 u32 blk; /* start block address of the extent */ 623 #ifdef CONFIG_F2FS_FS_COMPRESSION 624 unsigned int c_len; /* physical extent length of compressed blocks */ 625 #endif 626 }; 627 628 struct extent_node { 629 struct rb_node rb_node; /* rb node located in rb-tree */ 630 struct extent_info ei; /* extent info */ 631 struct list_head list; /* node in global extent list of sbi */ 632 struct extent_tree *et; /* extent tree pointer */ 633 }; 634 635 struct extent_tree { 636 nid_t ino; /* inode number */ 637 struct rb_root_cached root; /* root of extent info rb-tree */ 638 struct extent_node *cached_en; /* recently accessed extent node */ 639 struct extent_info largest; /* largested extent info */ 640 struct list_head list; /* to be used by sbi->zombie_list */ 641 rwlock_t lock; /* protect extent info rb-tree */ 642 atomic_t node_cnt; /* # of extent node in rb-tree*/ 643 bool largest_updated; /* largest extent updated */ 644 }; 645 646 /* 647 * This structure is taken from ext4_map_blocks. 648 * 649 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 650 */ 651 #define F2FS_MAP_NEW (1 << BH_New) 652 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 653 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 654 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 655 F2FS_MAP_UNWRITTEN) 656 657 struct f2fs_map_blocks { 658 struct block_device *m_bdev; /* for multi-device dio */ 659 block_t m_pblk; 660 block_t m_lblk; 661 unsigned int m_len; 662 unsigned int m_flags; 663 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 664 pgoff_t *m_next_extent; /* point to next possible extent */ 665 int m_seg_type; 666 bool m_may_create; /* indicate it is from write path */ 667 bool m_multidev_dio; /* indicate it allows multi-device dio */ 668 }; 669 670 /* for flag in get_data_block */ 671 enum { 672 F2FS_GET_BLOCK_DEFAULT, 673 F2FS_GET_BLOCK_FIEMAP, 674 F2FS_GET_BLOCK_BMAP, 675 F2FS_GET_BLOCK_DIO, 676 F2FS_GET_BLOCK_PRE_DIO, 677 F2FS_GET_BLOCK_PRE_AIO, 678 F2FS_GET_BLOCK_PRECACHE, 679 }; 680 681 /* 682 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 683 */ 684 #define FADVISE_COLD_BIT 0x01 685 #define FADVISE_LOST_PINO_BIT 0x02 686 #define FADVISE_ENCRYPT_BIT 0x04 687 #define FADVISE_ENC_NAME_BIT 0x08 688 #define FADVISE_KEEP_SIZE_BIT 0x10 689 #define FADVISE_HOT_BIT 0x20 690 #define FADVISE_VERITY_BIT 0x40 691 #define FADVISE_TRUNC_BIT 0x80 692 693 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 694 695 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 696 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 697 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 698 699 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 700 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 701 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 702 703 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 704 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 705 706 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 707 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 708 709 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 710 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 711 712 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 713 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 714 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 715 716 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 717 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 718 719 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 720 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 721 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 722 723 #define DEF_DIR_LEVEL 0 724 725 enum { 726 GC_FAILURE_PIN, 727 MAX_GC_FAILURE 728 }; 729 730 /* used for f2fs_inode_info->flags */ 731 enum { 732 FI_NEW_INODE, /* indicate newly allocated inode */ 733 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 734 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 735 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 736 FI_INC_LINK, /* need to increment i_nlink */ 737 FI_ACL_MODE, /* indicate acl mode */ 738 FI_NO_ALLOC, /* should not allocate any blocks */ 739 FI_FREE_NID, /* free allocated nide */ 740 FI_NO_EXTENT, /* not to use the extent cache */ 741 FI_INLINE_XATTR, /* used for inline xattr */ 742 FI_INLINE_DATA, /* used for inline data*/ 743 FI_INLINE_DENTRY, /* used for inline dentry */ 744 FI_APPEND_WRITE, /* inode has appended data */ 745 FI_UPDATE_WRITE, /* inode has in-place-update data */ 746 FI_NEED_IPU, /* used for ipu per file */ 747 FI_ATOMIC_FILE, /* indicate atomic file */ 748 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 749 FI_DROP_CACHE, /* drop dirty page cache */ 750 FI_DATA_EXIST, /* indicate data exists */ 751 FI_INLINE_DOTS, /* indicate inline dot dentries */ 752 FI_SKIP_WRITES, /* should skip data page writeback */ 753 FI_OPU_WRITE, /* used for opu per file */ 754 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 755 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 756 FI_HOT_DATA, /* indicate file is hot */ 757 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 758 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 759 FI_PIN_FILE, /* indicate file should not be gced */ 760 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 761 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 762 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 763 FI_MMAP_FILE, /* indicate file was mmapped */ 764 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 765 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 766 FI_ALIGNED_WRITE, /* enable aligned write */ 767 FI_COW_FILE, /* indicate COW file */ 768 FI_MAX, /* max flag, never be used */ 769 }; 770 771 struct f2fs_inode_info { 772 struct inode vfs_inode; /* serve a vfs inode */ 773 unsigned long i_flags; /* keep an inode flags for ioctl */ 774 unsigned char i_advise; /* use to give file attribute hints */ 775 unsigned char i_dir_level; /* use for dentry level for large dir */ 776 unsigned int i_current_depth; /* only for directory depth */ 777 /* for gc failure statistic */ 778 unsigned int i_gc_failures[MAX_GC_FAILURE]; 779 unsigned int i_pino; /* parent inode number */ 780 umode_t i_acl_mode; /* keep file acl mode temporarily */ 781 782 /* Use below internally in f2fs*/ 783 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 784 struct f2fs_rwsem i_sem; /* protect fi info */ 785 atomic_t dirty_pages; /* # of dirty pages */ 786 f2fs_hash_t chash; /* hash value of given file name */ 787 unsigned int clevel; /* maximum level of given file name */ 788 struct task_struct *task; /* lookup and create consistency */ 789 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 790 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 791 nid_t i_xattr_nid; /* node id that contains xattrs */ 792 loff_t last_disk_size; /* lastly written file size */ 793 spinlock_t i_size_lock; /* protect last_disk_size */ 794 795 #ifdef CONFIG_QUOTA 796 struct dquot *i_dquot[MAXQUOTAS]; 797 798 /* quota space reservation, managed internally by quota code */ 799 qsize_t i_reserved_quota; 800 #endif 801 struct list_head dirty_list; /* dirty list for dirs and files */ 802 struct list_head gdirty_list; /* linked in global dirty list */ 803 struct task_struct *atomic_write_task; /* store atomic write task */ 804 struct extent_tree *extent_tree; /* cached extent_tree entry */ 805 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 806 807 /* avoid racing between foreground op and gc */ 808 struct f2fs_rwsem i_gc_rwsem[2]; 809 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 810 811 int i_extra_isize; /* size of extra space located in i_addr */ 812 kprojid_t i_projid; /* id for project quota */ 813 int i_inline_xattr_size; /* inline xattr size */ 814 struct timespec64 i_crtime; /* inode creation time */ 815 struct timespec64 i_disk_time[4];/* inode disk times */ 816 817 /* for file compress */ 818 atomic_t i_compr_blocks; /* # of compressed blocks */ 819 unsigned char i_compress_algorithm; /* algorithm type */ 820 unsigned char i_log_cluster_size; /* log of cluster size */ 821 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 822 unsigned short i_compress_flag; /* compress flag */ 823 unsigned int i_cluster_size; /* cluster size */ 824 825 unsigned int atomic_write_cnt; 826 }; 827 828 static inline void get_extent_info(struct extent_info *ext, 829 struct f2fs_extent *i_ext) 830 { 831 ext->fofs = le32_to_cpu(i_ext->fofs); 832 ext->blk = le32_to_cpu(i_ext->blk); 833 ext->len = le32_to_cpu(i_ext->len); 834 } 835 836 static inline void set_raw_extent(struct extent_info *ext, 837 struct f2fs_extent *i_ext) 838 { 839 i_ext->fofs = cpu_to_le32(ext->fofs); 840 i_ext->blk = cpu_to_le32(ext->blk); 841 i_ext->len = cpu_to_le32(ext->len); 842 } 843 844 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 845 u32 blk, unsigned int len) 846 { 847 ei->fofs = fofs; 848 ei->blk = blk; 849 ei->len = len; 850 #ifdef CONFIG_F2FS_FS_COMPRESSION 851 ei->c_len = 0; 852 #endif 853 } 854 855 static inline bool __is_discard_mergeable(struct discard_info *back, 856 struct discard_info *front, unsigned int max_len) 857 { 858 return (back->lstart + back->len == front->lstart) && 859 (back->len + front->len <= max_len); 860 } 861 862 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 863 struct discard_info *back, unsigned int max_len) 864 { 865 return __is_discard_mergeable(back, cur, max_len); 866 } 867 868 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 869 struct discard_info *front, unsigned int max_len) 870 { 871 return __is_discard_mergeable(cur, front, max_len); 872 } 873 874 static inline bool __is_extent_mergeable(struct extent_info *back, 875 struct extent_info *front) 876 { 877 #ifdef CONFIG_F2FS_FS_COMPRESSION 878 if (back->c_len && back->len != back->c_len) 879 return false; 880 if (front->c_len && front->len != front->c_len) 881 return false; 882 #endif 883 return (back->fofs + back->len == front->fofs && 884 back->blk + back->len == front->blk); 885 } 886 887 static inline bool __is_back_mergeable(struct extent_info *cur, 888 struct extent_info *back) 889 { 890 return __is_extent_mergeable(back, cur); 891 } 892 893 static inline bool __is_front_mergeable(struct extent_info *cur, 894 struct extent_info *front) 895 { 896 return __is_extent_mergeable(cur, front); 897 } 898 899 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 900 static inline void __try_update_largest_extent(struct extent_tree *et, 901 struct extent_node *en) 902 { 903 if (en->ei.len > et->largest.len) { 904 et->largest = en->ei; 905 et->largest_updated = true; 906 } 907 } 908 909 /* 910 * For free nid management 911 */ 912 enum nid_state { 913 FREE_NID, /* newly added to free nid list */ 914 PREALLOC_NID, /* it is preallocated */ 915 MAX_NID_STATE, 916 }; 917 918 enum nat_state { 919 TOTAL_NAT, 920 DIRTY_NAT, 921 RECLAIMABLE_NAT, 922 MAX_NAT_STATE, 923 }; 924 925 struct f2fs_nm_info { 926 block_t nat_blkaddr; /* base disk address of NAT */ 927 nid_t max_nid; /* maximum possible node ids */ 928 nid_t available_nids; /* # of available node ids */ 929 nid_t next_scan_nid; /* the next nid to be scanned */ 930 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 931 unsigned int ram_thresh; /* control the memory footprint */ 932 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 933 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 934 935 /* NAT cache management */ 936 struct radix_tree_root nat_root;/* root of the nat entry cache */ 937 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 938 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 939 struct list_head nat_entries; /* cached nat entry list (clean) */ 940 spinlock_t nat_list_lock; /* protect clean nat entry list */ 941 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 942 unsigned int nat_blocks; /* # of nat blocks */ 943 944 /* free node ids management */ 945 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 946 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 947 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 948 spinlock_t nid_list_lock; /* protect nid lists ops */ 949 struct mutex build_lock; /* lock for build free nids */ 950 unsigned char **free_nid_bitmap; 951 unsigned char *nat_block_bitmap; 952 unsigned short *free_nid_count; /* free nid count of NAT block */ 953 954 /* for checkpoint */ 955 char *nat_bitmap; /* NAT bitmap pointer */ 956 957 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 958 unsigned char *nat_bits; /* NAT bits blocks */ 959 unsigned char *full_nat_bits; /* full NAT pages */ 960 unsigned char *empty_nat_bits; /* empty NAT pages */ 961 #ifdef CONFIG_F2FS_CHECK_FS 962 char *nat_bitmap_mir; /* NAT bitmap mirror */ 963 #endif 964 int bitmap_size; /* bitmap size */ 965 }; 966 967 /* 968 * this structure is used as one of function parameters. 969 * all the information are dedicated to a given direct node block determined 970 * by the data offset in a file. 971 */ 972 struct dnode_of_data { 973 struct inode *inode; /* vfs inode pointer */ 974 struct page *inode_page; /* its inode page, NULL is possible */ 975 struct page *node_page; /* cached direct node page */ 976 nid_t nid; /* node id of the direct node block */ 977 unsigned int ofs_in_node; /* data offset in the node page */ 978 bool inode_page_locked; /* inode page is locked or not */ 979 bool node_changed; /* is node block changed */ 980 char cur_level; /* level of hole node page */ 981 char max_level; /* level of current page located */ 982 block_t data_blkaddr; /* block address of the node block */ 983 }; 984 985 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 986 struct page *ipage, struct page *npage, nid_t nid) 987 { 988 memset(dn, 0, sizeof(*dn)); 989 dn->inode = inode; 990 dn->inode_page = ipage; 991 dn->node_page = npage; 992 dn->nid = nid; 993 } 994 995 /* 996 * For SIT manager 997 * 998 * By default, there are 6 active log areas across the whole main area. 999 * When considering hot and cold data separation to reduce cleaning overhead, 1000 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1001 * respectively. 1002 * In the current design, you should not change the numbers intentionally. 1003 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1004 * logs individually according to the underlying devices. (default: 6) 1005 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1006 * data and 8 for node logs. 1007 */ 1008 #define NR_CURSEG_DATA_TYPE (3) 1009 #define NR_CURSEG_NODE_TYPE (3) 1010 #define NR_CURSEG_INMEM_TYPE (2) 1011 #define NR_CURSEG_RO_TYPE (2) 1012 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1013 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1014 1015 enum { 1016 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1017 CURSEG_WARM_DATA, /* data blocks */ 1018 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1019 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1020 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1021 CURSEG_COLD_NODE, /* indirect node blocks */ 1022 NR_PERSISTENT_LOG, /* number of persistent log */ 1023 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1024 /* pinned file that needs consecutive block address */ 1025 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1026 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1027 }; 1028 1029 struct flush_cmd { 1030 struct completion wait; 1031 struct llist_node llnode; 1032 nid_t ino; 1033 int ret; 1034 }; 1035 1036 struct flush_cmd_control { 1037 struct task_struct *f2fs_issue_flush; /* flush thread */ 1038 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1039 atomic_t issued_flush; /* # of issued flushes */ 1040 atomic_t queued_flush; /* # of queued flushes */ 1041 struct llist_head issue_list; /* list for command issue */ 1042 struct llist_node *dispatch_list; /* list for command dispatch */ 1043 }; 1044 1045 struct f2fs_sm_info { 1046 struct sit_info *sit_info; /* whole segment information */ 1047 struct free_segmap_info *free_info; /* free segment information */ 1048 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1049 struct curseg_info *curseg_array; /* active segment information */ 1050 1051 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1052 1053 block_t seg0_blkaddr; /* block address of 0'th segment */ 1054 block_t main_blkaddr; /* start block address of main area */ 1055 block_t ssa_blkaddr; /* start block address of SSA area */ 1056 1057 unsigned int segment_count; /* total # of segments */ 1058 unsigned int main_segments; /* # of segments in main area */ 1059 unsigned int reserved_segments; /* # of reserved segments */ 1060 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1061 unsigned int ovp_segments; /* # of overprovision segments */ 1062 1063 /* a threshold to reclaim prefree segments */ 1064 unsigned int rec_prefree_segments; 1065 1066 /* for batched trimming */ 1067 unsigned int trim_sections; /* # of sections to trim */ 1068 1069 struct list_head sit_entry_set; /* sit entry set list */ 1070 1071 unsigned int ipu_policy; /* in-place-update policy */ 1072 unsigned int min_ipu_util; /* in-place-update threshold */ 1073 unsigned int min_fsync_blocks; /* threshold for fsync */ 1074 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1075 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1076 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1077 1078 /* for flush command control */ 1079 struct flush_cmd_control *fcc_info; 1080 1081 /* for discard command control */ 1082 struct discard_cmd_control *dcc_info; 1083 }; 1084 1085 /* 1086 * For superblock 1087 */ 1088 /* 1089 * COUNT_TYPE for monitoring 1090 * 1091 * f2fs monitors the number of several block types such as on-writeback, 1092 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1093 */ 1094 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1095 enum count_type { 1096 F2FS_DIRTY_DENTS, 1097 F2FS_DIRTY_DATA, 1098 F2FS_DIRTY_QDATA, 1099 F2FS_DIRTY_NODES, 1100 F2FS_DIRTY_META, 1101 F2FS_DIRTY_IMETA, 1102 F2FS_WB_CP_DATA, 1103 F2FS_WB_DATA, 1104 F2FS_RD_DATA, 1105 F2FS_RD_NODE, 1106 F2FS_RD_META, 1107 F2FS_DIO_WRITE, 1108 F2FS_DIO_READ, 1109 NR_COUNT_TYPE, 1110 }; 1111 1112 /* 1113 * The below are the page types of bios used in submit_bio(). 1114 * The available types are: 1115 * DATA User data pages. It operates as async mode. 1116 * NODE Node pages. It operates as async mode. 1117 * META FS metadata pages such as SIT, NAT, CP. 1118 * NR_PAGE_TYPE The number of page types. 1119 * META_FLUSH Make sure the previous pages are written 1120 * with waiting the bio's completion 1121 * ... Only can be used with META. 1122 */ 1123 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1124 enum page_type { 1125 DATA = 0, 1126 NODE = 1, /* should not change this */ 1127 META, 1128 NR_PAGE_TYPE, 1129 META_FLUSH, 1130 IPU, /* the below types are used by tracepoints only. */ 1131 OPU, 1132 }; 1133 1134 enum temp_type { 1135 HOT = 0, /* must be zero for meta bio */ 1136 WARM, 1137 COLD, 1138 NR_TEMP_TYPE, 1139 }; 1140 1141 enum need_lock_type { 1142 LOCK_REQ = 0, 1143 LOCK_DONE, 1144 LOCK_RETRY, 1145 }; 1146 1147 enum cp_reason_type { 1148 CP_NO_NEEDED, 1149 CP_NON_REGULAR, 1150 CP_COMPRESSED, 1151 CP_HARDLINK, 1152 CP_SB_NEED_CP, 1153 CP_WRONG_PINO, 1154 CP_NO_SPC_ROLL, 1155 CP_NODE_NEED_CP, 1156 CP_FASTBOOT_MODE, 1157 CP_SPEC_LOG_NUM, 1158 CP_RECOVER_DIR, 1159 }; 1160 1161 enum iostat_type { 1162 /* WRITE IO */ 1163 APP_DIRECT_IO, /* app direct write IOs */ 1164 APP_BUFFERED_IO, /* app buffered write IOs */ 1165 APP_WRITE_IO, /* app write IOs */ 1166 APP_MAPPED_IO, /* app mapped IOs */ 1167 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1168 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1169 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1170 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1171 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1172 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1173 FS_GC_DATA_IO, /* data IOs from forground gc */ 1174 FS_GC_NODE_IO, /* node IOs from forground gc */ 1175 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1176 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1177 FS_CP_META_IO, /* meta IOs from checkpoint */ 1178 1179 /* READ IO */ 1180 APP_DIRECT_READ_IO, /* app direct read IOs */ 1181 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1182 APP_READ_IO, /* app read IOs */ 1183 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1184 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1185 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1186 FS_DATA_READ_IO, /* data read IOs */ 1187 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1188 FS_CDATA_READ_IO, /* compressed data read IOs */ 1189 FS_NODE_READ_IO, /* node read IOs */ 1190 FS_META_READ_IO, /* meta read IOs */ 1191 1192 /* other */ 1193 FS_DISCARD, /* discard */ 1194 NR_IO_TYPE, 1195 }; 1196 1197 struct f2fs_io_info { 1198 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1199 nid_t ino; /* inode number */ 1200 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1201 enum temp_type temp; /* contains HOT/WARM/COLD */ 1202 enum req_op op; /* contains REQ_OP_ */ 1203 blk_opf_t op_flags; /* req_flag_bits */ 1204 block_t new_blkaddr; /* new block address to be written */ 1205 block_t old_blkaddr; /* old block address before Cow */ 1206 struct page *page; /* page to be written */ 1207 struct page *encrypted_page; /* encrypted page */ 1208 struct page *compressed_page; /* compressed page */ 1209 struct list_head list; /* serialize IOs */ 1210 bool submitted; /* indicate IO submission */ 1211 int need_lock; /* indicate we need to lock cp_rwsem */ 1212 bool in_list; /* indicate fio is in io_list */ 1213 bool is_por; /* indicate IO is from recovery or not */ 1214 bool retry; /* need to reallocate block address */ 1215 int compr_blocks; /* # of compressed block addresses */ 1216 bool encrypted; /* indicate file is encrypted */ 1217 bool post_read; /* require post read */ 1218 enum iostat_type io_type; /* io type */ 1219 struct writeback_control *io_wbc; /* writeback control */ 1220 struct bio **bio; /* bio for ipu */ 1221 sector_t *last_block; /* last block number in bio */ 1222 unsigned char version; /* version of the node */ 1223 }; 1224 1225 struct bio_entry { 1226 struct bio *bio; 1227 struct list_head list; 1228 }; 1229 1230 #define is_read_io(rw) ((rw) == READ) 1231 struct f2fs_bio_info { 1232 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1233 struct bio *bio; /* bios to merge */ 1234 sector_t last_block_in_bio; /* last block number */ 1235 struct f2fs_io_info fio; /* store buffered io info. */ 1236 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1237 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1238 struct list_head io_list; /* track fios */ 1239 struct list_head bio_list; /* bio entry list head */ 1240 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1241 }; 1242 1243 #define FDEV(i) (sbi->devs[i]) 1244 #define RDEV(i) (raw_super->devs[i]) 1245 struct f2fs_dev_info { 1246 struct block_device *bdev; 1247 char path[MAX_PATH_LEN]; 1248 unsigned int total_segments; 1249 block_t start_blk; 1250 block_t end_blk; 1251 #ifdef CONFIG_BLK_DEV_ZONED 1252 unsigned int nr_blkz; /* Total number of zones */ 1253 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1254 #endif 1255 }; 1256 1257 enum inode_type { 1258 DIR_INODE, /* for dirty dir inode */ 1259 FILE_INODE, /* for dirty regular/symlink inode */ 1260 DIRTY_META, /* for all dirtied inode metadata */ 1261 NR_INODE_TYPE, 1262 }; 1263 1264 /* for inner inode cache management */ 1265 struct inode_management { 1266 struct radix_tree_root ino_root; /* ino entry array */ 1267 spinlock_t ino_lock; /* for ino entry lock */ 1268 struct list_head ino_list; /* inode list head */ 1269 unsigned long ino_num; /* number of entries */ 1270 }; 1271 1272 /* for GC_AT */ 1273 struct atgc_management { 1274 bool atgc_enabled; /* ATGC is enabled or not */ 1275 struct rb_root_cached root; /* root of victim rb-tree */ 1276 struct list_head victim_list; /* linked with all victim entries */ 1277 unsigned int victim_count; /* victim count in rb-tree */ 1278 unsigned int candidate_ratio; /* candidate ratio */ 1279 unsigned int max_candidate_count; /* max candidate count */ 1280 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1281 unsigned long long age_threshold; /* age threshold */ 1282 }; 1283 1284 struct f2fs_gc_control { 1285 unsigned int victim_segno; /* target victim segment number */ 1286 int init_gc_type; /* FG_GC or BG_GC */ 1287 bool no_bg_gc; /* check the space and stop bg_gc */ 1288 bool should_migrate_blocks; /* should migrate blocks */ 1289 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1290 unsigned int nr_free_secs; /* # of free sections to do GC */ 1291 }; 1292 1293 /* For s_flag in struct f2fs_sb_info */ 1294 enum { 1295 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1296 SBI_IS_CLOSE, /* specify unmounting */ 1297 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1298 SBI_POR_DOING, /* recovery is doing or not */ 1299 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1300 SBI_NEED_CP, /* need to checkpoint */ 1301 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1302 SBI_IS_RECOVERED, /* recovered orphan/data */ 1303 SBI_CP_DISABLED, /* CP was disabled last mount */ 1304 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1305 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1306 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1307 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1308 SBI_IS_RESIZEFS, /* resizefs is in process */ 1309 SBI_IS_FREEZING, /* freezefs is in process */ 1310 }; 1311 1312 enum { 1313 CP_TIME, 1314 REQ_TIME, 1315 DISCARD_TIME, 1316 GC_TIME, 1317 DISABLE_TIME, 1318 UMOUNT_DISCARD_TIMEOUT, 1319 MAX_TIME, 1320 }; 1321 1322 enum { 1323 GC_NORMAL, 1324 GC_IDLE_CB, 1325 GC_IDLE_GREEDY, 1326 GC_IDLE_AT, 1327 GC_URGENT_HIGH, 1328 GC_URGENT_LOW, 1329 GC_URGENT_MID, 1330 MAX_GC_MODE, 1331 }; 1332 1333 enum { 1334 BGGC_MODE_ON, /* background gc is on */ 1335 BGGC_MODE_OFF, /* background gc is off */ 1336 BGGC_MODE_SYNC, /* 1337 * background gc is on, migrating blocks 1338 * like foreground gc 1339 */ 1340 }; 1341 1342 enum { 1343 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1344 FS_MODE_LFS, /* use lfs allocation only */ 1345 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1346 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1347 }; 1348 1349 enum { 1350 ALLOC_MODE_DEFAULT, /* stay default */ 1351 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1352 }; 1353 1354 enum fsync_mode { 1355 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1356 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1357 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1358 }; 1359 1360 enum { 1361 COMPR_MODE_FS, /* 1362 * automatically compress compression 1363 * enabled files 1364 */ 1365 COMPR_MODE_USER, /* 1366 * automatical compression is disabled. 1367 * user can control the file compression 1368 * using ioctls 1369 */ 1370 }; 1371 1372 enum { 1373 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1374 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1375 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1376 }; 1377 1378 enum { 1379 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1380 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1381 }; 1382 1383 1384 1385 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1386 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1387 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1388 1389 /* 1390 * Layout of f2fs page.private: 1391 * 1392 * Layout A: lowest bit should be 1 1393 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1394 * bit 0 PAGE_PRIVATE_NOT_POINTER 1395 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1396 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1397 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1398 * bit 4 PAGE_PRIVATE_INLINE_INODE 1399 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1400 * bit 6- f2fs private data 1401 * 1402 * Layout B: lowest bit should be 0 1403 * page.private is a wrapped pointer. 1404 */ 1405 enum { 1406 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1407 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1408 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1409 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1410 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1411 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1412 PAGE_PRIVATE_MAX 1413 }; 1414 1415 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1416 static inline bool page_private_##name(struct page *page) \ 1417 { \ 1418 return PagePrivate(page) && \ 1419 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1420 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1421 } 1422 1423 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1424 static inline void set_page_private_##name(struct page *page) \ 1425 { \ 1426 if (!PagePrivate(page)) { \ 1427 get_page(page); \ 1428 SetPagePrivate(page); \ 1429 set_page_private(page, 0); \ 1430 } \ 1431 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1432 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1433 } 1434 1435 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1436 static inline void clear_page_private_##name(struct page *page) \ 1437 { \ 1438 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1439 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1440 set_page_private(page, 0); \ 1441 if (PagePrivate(page)) { \ 1442 ClearPagePrivate(page); \ 1443 put_page(page); \ 1444 }\ 1445 } \ 1446 } 1447 1448 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1449 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1450 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1451 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1452 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1453 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1454 1455 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1456 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1457 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1458 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1459 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1460 1461 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1462 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1463 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1464 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1465 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1466 1467 static inline unsigned long get_page_private_data(struct page *page) 1468 { 1469 unsigned long data = page_private(page); 1470 1471 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1472 return 0; 1473 return data >> PAGE_PRIVATE_MAX; 1474 } 1475 1476 static inline void set_page_private_data(struct page *page, unsigned long data) 1477 { 1478 if (!PagePrivate(page)) { 1479 get_page(page); 1480 SetPagePrivate(page); 1481 set_page_private(page, 0); 1482 } 1483 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1484 page_private(page) |= data << PAGE_PRIVATE_MAX; 1485 } 1486 1487 static inline void clear_page_private_data(struct page *page) 1488 { 1489 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1490 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1491 set_page_private(page, 0); 1492 if (PagePrivate(page)) { 1493 ClearPagePrivate(page); 1494 put_page(page); 1495 } 1496 } 1497 } 1498 1499 /* For compression */ 1500 enum compress_algorithm_type { 1501 COMPRESS_LZO, 1502 COMPRESS_LZ4, 1503 COMPRESS_ZSTD, 1504 COMPRESS_LZORLE, 1505 COMPRESS_MAX, 1506 }; 1507 1508 enum compress_flag { 1509 COMPRESS_CHKSUM, 1510 COMPRESS_MAX_FLAG, 1511 }; 1512 1513 #define COMPRESS_WATERMARK 20 1514 #define COMPRESS_PERCENT 20 1515 1516 #define COMPRESS_DATA_RESERVED_SIZE 4 1517 struct compress_data { 1518 __le32 clen; /* compressed data size */ 1519 __le32 chksum; /* compressed data chksum */ 1520 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1521 u8 cdata[]; /* compressed data */ 1522 }; 1523 1524 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1525 1526 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1527 1528 #define COMPRESS_LEVEL_OFFSET 8 1529 1530 /* compress context */ 1531 struct compress_ctx { 1532 struct inode *inode; /* inode the context belong to */ 1533 pgoff_t cluster_idx; /* cluster index number */ 1534 unsigned int cluster_size; /* page count in cluster */ 1535 unsigned int log_cluster_size; /* log of cluster size */ 1536 struct page **rpages; /* pages store raw data in cluster */ 1537 unsigned int nr_rpages; /* total page number in rpages */ 1538 struct page **cpages; /* pages store compressed data in cluster */ 1539 unsigned int nr_cpages; /* total page number in cpages */ 1540 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1541 void *rbuf; /* virtual mapped address on rpages */ 1542 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1543 size_t rlen; /* valid data length in rbuf */ 1544 size_t clen; /* valid data length in cbuf */ 1545 void *private; /* payload buffer for specified compression algorithm */ 1546 void *private2; /* extra payload buffer */ 1547 }; 1548 1549 /* compress context for write IO path */ 1550 struct compress_io_ctx { 1551 u32 magic; /* magic number to indicate page is compressed */ 1552 struct inode *inode; /* inode the context belong to */ 1553 struct page **rpages; /* pages store raw data in cluster */ 1554 unsigned int nr_rpages; /* total page number in rpages */ 1555 atomic_t pending_pages; /* in-flight compressed page count */ 1556 }; 1557 1558 /* Context for decompressing one cluster on the read IO path */ 1559 struct decompress_io_ctx { 1560 u32 magic; /* magic number to indicate page is compressed */ 1561 struct inode *inode; /* inode the context belong to */ 1562 pgoff_t cluster_idx; /* cluster index number */ 1563 unsigned int cluster_size; /* page count in cluster */ 1564 unsigned int log_cluster_size; /* log of cluster size */ 1565 struct page **rpages; /* pages store raw data in cluster */ 1566 unsigned int nr_rpages; /* total page number in rpages */ 1567 struct page **cpages; /* pages store compressed data in cluster */ 1568 unsigned int nr_cpages; /* total page number in cpages */ 1569 struct page **tpages; /* temp pages to pad holes in cluster */ 1570 void *rbuf; /* virtual mapped address on rpages */ 1571 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1572 size_t rlen; /* valid data length in rbuf */ 1573 size_t clen; /* valid data length in cbuf */ 1574 1575 /* 1576 * The number of compressed pages remaining to be read in this cluster. 1577 * This is initially nr_cpages. It is decremented by 1 each time a page 1578 * has been read (or failed to be read). When it reaches 0, the cluster 1579 * is decompressed (or an error is reported). 1580 * 1581 * If an error occurs before all the pages have been submitted for I/O, 1582 * then this will never reach 0. In this case the I/O submitter is 1583 * responsible for calling f2fs_decompress_end_io() instead. 1584 */ 1585 atomic_t remaining_pages; 1586 1587 /* 1588 * Number of references to this decompress_io_ctx. 1589 * 1590 * One reference is held for I/O completion. This reference is dropped 1591 * after the pagecache pages are updated and unlocked -- either after 1592 * decompression (and verity if enabled), or after an error. 1593 * 1594 * In addition, each compressed page holds a reference while it is in a 1595 * bio. These references are necessary prevent compressed pages from 1596 * being freed while they are still in a bio. 1597 */ 1598 refcount_t refcnt; 1599 1600 bool failed; /* IO error occurred before decompression? */ 1601 bool need_verity; /* need fs-verity verification after decompression? */ 1602 void *private; /* payload buffer for specified decompression algorithm */ 1603 void *private2; /* extra payload buffer */ 1604 struct work_struct verity_work; /* work to verify the decompressed pages */ 1605 struct work_struct free_work; /* work for late free this structure itself */ 1606 }; 1607 1608 #define NULL_CLUSTER ((unsigned int)(~0)) 1609 #define MIN_COMPRESS_LOG_SIZE 2 1610 #define MAX_COMPRESS_LOG_SIZE 8 1611 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1612 1613 struct f2fs_sb_info { 1614 struct super_block *sb; /* pointer to VFS super block */ 1615 struct proc_dir_entry *s_proc; /* proc entry */ 1616 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1617 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1618 int valid_super_block; /* valid super block no */ 1619 unsigned long s_flag; /* flags for sbi */ 1620 struct mutex writepages; /* mutex for writepages() */ 1621 1622 #ifdef CONFIG_BLK_DEV_ZONED 1623 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1624 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1625 #endif 1626 1627 /* for node-related operations */ 1628 struct f2fs_nm_info *nm_info; /* node manager */ 1629 struct inode *node_inode; /* cache node blocks */ 1630 1631 /* for segment-related operations */ 1632 struct f2fs_sm_info *sm_info; /* segment manager */ 1633 1634 /* for bio operations */ 1635 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1636 /* keep migration IO order for LFS mode */ 1637 struct f2fs_rwsem io_order_lock; 1638 mempool_t *write_io_dummy; /* Dummy pages */ 1639 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1640 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1641 1642 /* for checkpoint */ 1643 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1644 int cur_cp_pack; /* remain current cp pack */ 1645 spinlock_t cp_lock; /* for flag in ckpt */ 1646 struct inode *meta_inode; /* cache meta blocks */ 1647 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1648 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1649 struct f2fs_rwsem node_write; /* locking node writes */ 1650 struct f2fs_rwsem node_change; /* locking node change */ 1651 wait_queue_head_t cp_wait; 1652 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1653 long interval_time[MAX_TIME]; /* to store thresholds */ 1654 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1655 1656 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1657 1658 spinlock_t fsync_node_lock; /* for node entry lock */ 1659 struct list_head fsync_node_list; /* node list head */ 1660 unsigned int fsync_seg_id; /* sequence id */ 1661 unsigned int fsync_node_num; /* number of node entries */ 1662 1663 /* for orphan inode, use 0'th array */ 1664 unsigned int max_orphans; /* max orphan inodes */ 1665 1666 /* for inode management */ 1667 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1668 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1669 struct mutex flush_lock; /* for flush exclusion */ 1670 1671 /* for extent tree cache */ 1672 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1673 struct mutex extent_tree_lock; /* locking extent radix tree */ 1674 struct list_head extent_list; /* lru list for shrinker */ 1675 spinlock_t extent_lock; /* locking extent lru list */ 1676 atomic_t total_ext_tree; /* extent tree count */ 1677 struct list_head zombie_list; /* extent zombie tree list */ 1678 atomic_t total_zombie_tree; /* extent zombie tree count */ 1679 atomic_t total_ext_node; /* extent info count */ 1680 1681 /* basic filesystem units */ 1682 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1683 unsigned int log_blocksize; /* log2 block size */ 1684 unsigned int blocksize; /* block size */ 1685 unsigned int root_ino_num; /* root inode number*/ 1686 unsigned int node_ino_num; /* node inode number*/ 1687 unsigned int meta_ino_num; /* meta inode number*/ 1688 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1689 unsigned int blocks_per_seg; /* blocks per segment */ 1690 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1691 unsigned int segs_per_sec; /* segments per section */ 1692 unsigned int secs_per_zone; /* sections per zone */ 1693 unsigned int total_sections; /* total section count */ 1694 unsigned int total_node_count; /* total node block count */ 1695 unsigned int total_valid_node_count; /* valid node block count */ 1696 int dir_level; /* directory level */ 1697 int readdir_ra; /* readahead inode in readdir */ 1698 u64 max_io_bytes; /* max io bytes to merge IOs */ 1699 1700 block_t user_block_count; /* # of user blocks */ 1701 block_t total_valid_block_count; /* # of valid blocks */ 1702 block_t discard_blks; /* discard command candidats */ 1703 block_t last_valid_block_count; /* for recovery */ 1704 block_t reserved_blocks; /* configurable reserved blocks */ 1705 block_t current_reserved_blocks; /* current reserved blocks */ 1706 1707 /* Additional tracking for no checkpoint mode */ 1708 block_t unusable_block_count; /* # of blocks saved by last cp */ 1709 1710 unsigned int nquota_files; /* # of quota sysfile */ 1711 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1712 1713 /* # of pages, see count_type */ 1714 atomic_t nr_pages[NR_COUNT_TYPE]; 1715 /* # of allocated blocks */ 1716 struct percpu_counter alloc_valid_block_count; 1717 /* # of node block writes as roll forward recovery */ 1718 struct percpu_counter rf_node_block_count; 1719 1720 /* writeback control */ 1721 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1722 1723 /* valid inode count */ 1724 struct percpu_counter total_valid_inode_count; 1725 1726 struct f2fs_mount_info mount_opt; /* mount options */ 1727 1728 /* for cleaning operations */ 1729 struct f2fs_rwsem gc_lock; /* 1730 * semaphore for GC, avoid 1731 * race between GC and GC or CP 1732 */ 1733 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1734 struct atgc_management am; /* atgc management */ 1735 unsigned int cur_victim_sec; /* current victim section num */ 1736 unsigned int gc_mode; /* current GC state */ 1737 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1738 spinlock_t gc_urgent_high_lock; 1739 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1740 1741 /* for skip statistic */ 1742 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1743 1744 /* threshold for gc trials on pinned files */ 1745 u64 gc_pin_file_threshold; 1746 struct f2fs_rwsem pin_sem; 1747 1748 /* maximum # of trials to find a victim segment for SSR and GC */ 1749 unsigned int max_victim_search; 1750 /* migration granularity of garbage collection, unit: segment */ 1751 unsigned int migration_granularity; 1752 1753 /* 1754 * for stat information. 1755 * one is for the LFS mode, and the other is for the SSR mode. 1756 */ 1757 #ifdef CONFIG_F2FS_STAT_FS 1758 struct f2fs_stat_info *stat_info; /* FS status information */ 1759 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1760 unsigned int segment_count[2]; /* # of allocated segments */ 1761 unsigned int block_count[2]; /* # of allocated blocks */ 1762 atomic_t inplace_count; /* # of inplace update */ 1763 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1764 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1765 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1766 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1767 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1768 atomic_t inline_inode; /* # of inline_data inodes */ 1769 atomic_t inline_dir; /* # of inline_dentry inodes */ 1770 atomic_t compr_inode; /* # of compressed inodes */ 1771 atomic64_t compr_blocks; /* # of compressed blocks */ 1772 atomic_t swapfile_inode; /* # of swapfile inodes */ 1773 atomic_t atomic_files; /* # of opened atomic file */ 1774 atomic_t max_aw_cnt; /* max # of atomic writes */ 1775 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1776 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1777 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1778 #endif 1779 spinlock_t stat_lock; /* lock for stat operations */ 1780 1781 /* to attach REQ_META|REQ_FUA flags */ 1782 unsigned int data_io_flag; 1783 unsigned int node_io_flag; 1784 1785 /* For sysfs support */ 1786 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1787 struct completion s_kobj_unregister; 1788 1789 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1790 struct completion s_stat_kobj_unregister; 1791 1792 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1793 struct completion s_feature_list_kobj_unregister; 1794 1795 /* For shrinker support */ 1796 struct list_head s_list; 1797 struct mutex umount_mutex; 1798 unsigned int shrinker_run_no; 1799 1800 /* For multi devices */ 1801 int s_ndevs; /* number of devices */ 1802 struct f2fs_dev_info *devs; /* for device list */ 1803 unsigned int dirty_device; /* for checkpoint data flush */ 1804 spinlock_t dev_lock; /* protect dirty_device */ 1805 bool aligned_blksize; /* all devices has the same logical blksize */ 1806 1807 /* For write statistics */ 1808 u64 sectors_written_start; 1809 u64 kbytes_written; 1810 1811 /* Reference to checksum algorithm driver via cryptoapi */ 1812 struct crypto_shash *s_chksum_driver; 1813 1814 /* Precomputed FS UUID checksum for seeding other checksums */ 1815 __u32 s_chksum_seed; 1816 1817 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1818 1819 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1820 spinlock_t error_lock; /* protect errors array */ 1821 bool error_dirty; /* errors of sb is dirty */ 1822 1823 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1824 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1825 1826 /* For reclaimed segs statistics per each GC mode */ 1827 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1828 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1829 1830 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1831 1832 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1833 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1834 1835 /* For atomic write statistics */ 1836 atomic64_t current_atomic_write; 1837 s64 peak_atomic_write; 1838 u64 committed_atomic_block; 1839 u64 revoked_atomic_block; 1840 1841 #ifdef CONFIG_F2FS_FS_COMPRESSION 1842 struct kmem_cache *page_array_slab; /* page array entry */ 1843 unsigned int page_array_slab_size; /* default page array slab size */ 1844 1845 /* For runtime compression statistics */ 1846 u64 compr_written_block; 1847 u64 compr_saved_block; 1848 u32 compr_new_inode; 1849 1850 /* For compressed block cache */ 1851 struct inode *compress_inode; /* cache compressed blocks */ 1852 unsigned int compress_percent; /* cache page percentage */ 1853 unsigned int compress_watermark; /* cache page watermark */ 1854 atomic_t compress_page_hit; /* cache hit count */ 1855 #endif 1856 1857 #ifdef CONFIG_F2FS_IOSTAT 1858 /* For app/fs IO statistics */ 1859 spinlock_t iostat_lock; 1860 unsigned long long rw_iostat[NR_IO_TYPE]; 1861 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1862 bool iostat_enable; 1863 unsigned long iostat_next_period; 1864 unsigned int iostat_period_ms; 1865 1866 /* For io latency related statistics info in one iostat period */ 1867 spinlock_t iostat_lat_lock; 1868 struct iostat_lat_info *iostat_io_lat; 1869 #endif 1870 }; 1871 1872 #ifdef CONFIG_F2FS_FAULT_INJECTION 1873 #define f2fs_show_injection_info(sbi, type) \ 1874 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1875 KERN_INFO, sbi->sb->s_id, \ 1876 f2fs_fault_name[type], \ 1877 __func__, __builtin_return_address(0)) 1878 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1879 { 1880 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1881 1882 if (!ffi->inject_rate) 1883 return false; 1884 1885 if (!IS_FAULT_SET(ffi, type)) 1886 return false; 1887 1888 atomic_inc(&ffi->inject_ops); 1889 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1890 atomic_set(&ffi->inject_ops, 0); 1891 return true; 1892 } 1893 return false; 1894 } 1895 #else 1896 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1897 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1898 { 1899 return false; 1900 } 1901 #endif 1902 1903 /* 1904 * Test if the mounted volume is a multi-device volume. 1905 * - For a single regular disk volume, sbi->s_ndevs is 0. 1906 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1907 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1908 */ 1909 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1910 { 1911 return sbi->s_ndevs > 1; 1912 } 1913 1914 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1915 { 1916 unsigned long now = jiffies; 1917 1918 sbi->last_time[type] = now; 1919 1920 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1921 if (type == REQ_TIME) { 1922 sbi->last_time[DISCARD_TIME] = now; 1923 sbi->last_time[GC_TIME] = now; 1924 } 1925 } 1926 1927 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1928 { 1929 unsigned long interval = sbi->interval_time[type] * HZ; 1930 1931 return time_after(jiffies, sbi->last_time[type] + interval); 1932 } 1933 1934 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1935 int type) 1936 { 1937 unsigned long interval = sbi->interval_time[type] * HZ; 1938 unsigned int wait_ms = 0; 1939 long delta; 1940 1941 delta = (sbi->last_time[type] + interval) - jiffies; 1942 if (delta > 0) 1943 wait_ms = jiffies_to_msecs(delta); 1944 1945 return wait_ms; 1946 } 1947 1948 /* 1949 * Inline functions 1950 */ 1951 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1952 const void *address, unsigned int length) 1953 { 1954 struct { 1955 struct shash_desc shash; 1956 char ctx[4]; 1957 } desc; 1958 int err; 1959 1960 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1961 1962 desc.shash.tfm = sbi->s_chksum_driver; 1963 *(u32 *)desc.ctx = crc; 1964 1965 err = crypto_shash_update(&desc.shash, address, length); 1966 BUG_ON(err); 1967 1968 return *(u32 *)desc.ctx; 1969 } 1970 1971 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1972 unsigned int length) 1973 { 1974 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1975 } 1976 1977 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1978 void *buf, size_t buf_size) 1979 { 1980 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1981 } 1982 1983 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1984 const void *address, unsigned int length) 1985 { 1986 return __f2fs_crc32(sbi, crc, address, length); 1987 } 1988 1989 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1990 { 1991 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1992 } 1993 1994 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1995 { 1996 return sb->s_fs_info; 1997 } 1998 1999 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2000 { 2001 return F2FS_SB(inode->i_sb); 2002 } 2003 2004 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2005 { 2006 return F2FS_I_SB(mapping->host); 2007 } 2008 2009 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 2010 { 2011 return F2FS_M_SB(page_file_mapping(page)); 2012 } 2013 2014 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2015 { 2016 return (struct f2fs_super_block *)(sbi->raw_super); 2017 } 2018 2019 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2020 { 2021 return (struct f2fs_checkpoint *)(sbi->ckpt); 2022 } 2023 2024 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2025 { 2026 return (struct f2fs_node *)page_address(page); 2027 } 2028 2029 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2030 { 2031 return &((struct f2fs_node *)page_address(page))->i; 2032 } 2033 2034 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2035 { 2036 return (struct f2fs_nm_info *)(sbi->nm_info); 2037 } 2038 2039 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2040 { 2041 return (struct f2fs_sm_info *)(sbi->sm_info); 2042 } 2043 2044 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2045 { 2046 return (struct sit_info *)(SM_I(sbi)->sit_info); 2047 } 2048 2049 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2050 { 2051 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2052 } 2053 2054 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2055 { 2056 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2057 } 2058 2059 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2060 { 2061 return sbi->meta_inode->i_mapping; 2062 } 2063 2064 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2065 { 2066 return sbi->node_inode->i_mapping; 2067 } 2068 2069 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2070 { 2071 return test_bit(type, &sbi->s_flag); 2072 } 2073 2074 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2075 { 2076 set_bit(type, &sbi->s_flag); 2077 } 2078 2079 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2080 { 2081 clear_bit(type, &sbi->s_flag); 2082 } 2083 2084 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2085 { 2086 return le64_to_cpu(cp->checkpoint_ver); 2087 } 2088 2089 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2090 { 2091 if (type < F2FS_MAX_QUOTAS) 2092 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2093 return 0; 2094 } 2095 2096 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2097 { 2098 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2099 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2100 } 2101 2102 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2103 { 2104 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2105 2106 return ckpt_flags & f; 2107 } 2108 2109 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2110 { 2111 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2112 } 2113 2114 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2115 { 2116 unsigned int ckpt_flags; 2117 2118 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2119 ckpt_flags |= f; 2120 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2121 } 2122 2123 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2124 { 2125 unsigned long flags; 2126 2127 spin_lock_irqsave(&sbi->cp_lock, flags); 2128 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2129 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2130 } 2131 2132 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2133 { 2134 unsigned int ckpt_flags; 2135 2136 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2137 ckpt_flags &= (~f); 2138 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2139 } 2140 2141 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2142 { 2143 unsigned long flags; 2144 2145 spin_lock_irqsave(&sbi->cp_lock, flags); 2146 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2147 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2148 } 2149 2150 #define init_f2fs_rwsem(sem) \ 2151 do { \ 2152 static struct lock_class_key __key; \ 2153 \ 2154 __init_f2fs_rwsem((sem), #sem, &__key); \ 2155 } while (0) 2156 2157 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2158 const char *sem_name, struct lock_class_key *key) 2159 { 2160 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2161 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2162 init_waitqueue_head(&sem->read_waiters); 2163 #endif 2164 } 2165 2166 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2167 { 2168 return rwsem_is_locked(&sem->internal_rwsem); 2169 } 2170 2171 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2172 { 2173 return rwsem_is_contended(&sem->internal_rwsem); 2174 } 2175 2176 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2177 { 2178 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2179 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2180 #else 2181 down_read(&sem->internal_rwsem); 2182 #endif 2183 } 2184 2185 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2186 { 2187 return down_read_trylock(&sem->internal_rwsem); 2188 } 2189 2190 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2191 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2192 { 2193 down_read_nested(&sem->internal_rwsem, subclass); 2194 } 2195 #else 2196 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2197 #endif 2198 2199 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2200 { 2201 up_read(&sem->internal_rwsem); 2202 } 2203 2204 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2205 { 2206 down_write(&sem->internal_rwsem); 2207 } 2208 2209 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2210 { 2211 return down_write_trylock(&sem->internal_rwsem); 2212 } 2213 2214 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2215 { 2216 up_write(&sem->internal_rwsem); 2217 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2218 wake_up_all(&sem->read_waiters); 2219 #endif 2220 } 2221 2222 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2223 { 2224 f2fs_down_read(&sbi->cp_rwsem); 2225 } 2226 2227 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2228 { 2229 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2230 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2231 return 0; 2232 } 2233 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2234 } 2235 2236 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2237 { 2238 f2fs_up_read(&sbi->cp_rwsem); 2239 } 2240 2241 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2242 { 2243 f2fs_down_write(&sbi->cp_rwsem); 2244 } 2245 2246 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2247 { 2248 f2fs_up_write(&sbi->cp_rwsem); 2249 } 2250 2251 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2252 { 2253 int reason = CP_SYNC; 2254 2255 if (test_opt(sbi, FASTBOOT)) 2256 reason = CP_FASTBOOT; 2257 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2258 reason = CP_UMOUNT; 2259 return reason; 2260 } 2261 2262 static inline bool __remain_node_summaries(int reason) 2263 { 2264 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2265 } 2266 2267 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2268 { 2269 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2270 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2271 } 2272 2273 /* 2274 * Check whether the inode has blocks or not 2275 */ 2276 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2277 { 2278 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2279 2280 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2281 } 2282 2283 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2284 { 2285 return ofs == XATTR_NODE_OFFSET; 2286 } 2287 2288 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2289 struct inode *inode, bool cap) 2290 { 2291 if (!inode) 2292 return true; 2293 if (!test_opt(sbi, RESERVE_ROOT)) 2294 return false; 2295 if (IS_NOQUOTA(inode)) 2296 return true; 2297 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2298 return true; 2299 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2300 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2301 return true; 2302 if (cap && capable(CAP_SYS_RESOURCE)) 2303 return true; 2304 return false; 2305 } 2306 2307 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2308 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2309 struct inode *inode, blkcnt_t *count) 2310 { 2311 blkcnt_t diff = 0, release = 0; 2312 block_t avail_user_block_count; 2313 int ret; 2314 2315 ret = dquot_reserve_block(inode, *count); 2316 if (ret) 2317 return ret; 2318 2319 if (time_to_inject(sbi, FAULT_BLOCK)) { 2320 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2321 release = *count; 2322 goto release_quota; 2323 } 2324 2325 /* 2326 * let's increase this in prior to actual block count change in order 2327 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2328 */ 2329 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2330 2331 spin_lock(&sbi->stat_lock); 2332 sbi->total_valid_block_count += (block_t)(*count); 2333 avail_user_block_count = sbi->user_block_count - 2334 sbi->current_reserved_blocks; 2335 2336 if (!__allow_reserved_blocks(sbi, inode, true)) 2337 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2338 2339 if (F2FS_IO_ALIGNED(sbi)) 2340 avail_user_block_count -= sbi->blocks_per_seg * 2341 SM_I(sbi)->additional_reserved_segments; 2342 2343 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2344 if (avail_user_block_count > sbi->unusable_block_count) 2345 avail_user_block_count -= sbi->unusable_block_count; 2346 else 2347 avail_user_block_count = 0; 2348 } 2349 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2350 diff = sbi->total_valid_block_count - avail_user_block_count; 2351 if (diff > *count) 2352 diff = *count; 2353 *count -= diff; 2354 release = diff; 2355 sbi->total_valid_block_count -= diff; 2356 if (!*count) { 2357 spin_unlock(&sbi->stat_lock); 2358 goto enospc; 2359 } 2360 } 2361 spin_unlock(&sbi->stat_lock); 2362 2363 if (unlikely(release)) { 2364 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2365 dquot_release_reservation_block(inode, release); 2366 } 2367 f2fs_i_blocks_write(inode, *count, true, true); 2368 return 0; 2369 2370 enospc: 2371 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2372 release_quota: 2373 dquot_release_reservation_block(inode, release); 2374 return -ENOSPC; 2375 } 2376 2377 __printf(2, 3) 2378 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2379 2380 #define f2fs_err(sbi, fmt, ...) \ 2381 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2382 #define f2fs_warn(sbi, fmt, ...) \ 2383 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2384 #define f2fs_notice(sbi, fmt, ...) \ 2385 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2386 #define f2fs_info(sbi, fmt, ...) \ 2387 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2388 #define f2fs_debug(sbi, fmt, ...) \ 2389 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2390 2391 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2392 struct inode *inode, 2393 block_t count) 2394 { 2395 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2396 2397 spin_lock(&sbi->stat_lock); 2398 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2399 sbi->total_valid_block_count -= (block_t)count; 2400 if (sbi->reserved_blocks && 2401 sbi->current_reserved_blocks < sbi->reserved_blocks) 2402 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2403 sbi->current_reserved_blocks + count); 2404 spin_unlock(&sbi->stat_lock); 2405 if (unlikely(inode->i_blocks < sectors)) { 2406 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2407 inode->i_ino, 2408 (unsigned long long)inode->i_blocks, 2409 (unsigned long long)sectors); 2410 set_sbi_flag(sbi, SBI_NEED_FSCK); 2411 return; 2412 } 2413 f2fs_i_blocks_write(inode, count, false, true); 2414 } 2415 2416 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2417 { 2418 atomic_inc(&sbi->nr_pages[count_type]); 2419 2420 if (count_type == F2FS_DIRTY_DENTS || 2421 count_type == F2FS_DIRTY_NODES || 2422 count_type == F2FS_DIRTY_META || 2423 count_type == F2FS_DIRTY_QDATA || 2424 count_type == F2FS_DIRTY_IMETA) 2425 set_sbi_flag(sbi, SBI_IS_DIRTY); 2426 } 2427 2428 static inline void inode_inc_dirty_pages(struct inode *inode) 2429 { 2430 atomic_inc(&F2FS_I(inode)->dirty_pages); 2431 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2432 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2433 if (IS_NOQUOTA(inode)) 2434 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2435 } 2436 2437 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2438 { 2439 atomic_dec(&sbi->nr_pages[count_type]); 2440 } 2441 2442 static inline void inode_dec_dirty_pages(struct inode *inode) 2443 { 2444 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2445 !S_ISLNK(inode->i_mode)) 2446 return; 2447 2448 atomic_dec(&F2FS_I(inode)->dirty_pages); 2449 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2450 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2451 if (IS_NOQUOTA(inode)) 2452 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2453 } 2454 2455 static inline void inc_atomic_write_cnt(struct inode *inode) 2456 { 2457 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2458 struct f2fs_inode_info *fi = F2FS_I(inode); 2459 u64 current_write; 2460 2461 fi->atomic_write_cnt++; 2462 atomic64_inc(&sbi->current_atomic_write); 2463 current_write = atomic64_read(&sbi->current_atomic_write); 2464 if (current_write > sbi->peak_atomic_write) 2465 sbi->peak_atomic_write = current_write; 2466 } 2467 2468 static inline void release_atomic_write_cnt(struct inode *inode) 2469 { 2470 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2471 struct f2fs_inode_info *fi = F2FS_I(inode); 2472 2473 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2474 fi->atomic_write_cnt = 0; 2475 } 2476 2477 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2478 { 2479 return atomic_read(&sbi->nr_pages[count_type]); 2480 } 2481 2482 static inline int get_dirty_pages(struct inode *inode) 2483 { 2484 return atomic_read(&F2FS_I(inode)->dirty_pages); 2485 } 2486 2487 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2488 { 2489 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2490 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2491 sbi->log_blocks_per_seg; 2492 2493 return segs / sbi->segs_per_sec; 2494 } 2495 2496 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2497 { 2498 return sbi->total_valid_block_count; 2499 } 2500 2501 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2502 { 2503 return sbi->discard_blks; 2504 } 2505 2506 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2507 { 2508 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2509 2510 /* return NAT or SIT bitmap */ 2511 if (flag == NAT_BITMAP) 2512 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2513 else if (flag == SIT_BITMAP) 2514 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2515 2516 return 0; 2517 } 2518 2519 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2520 { 2521 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2522 } 2523 2524 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2525 { 2526 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2527 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2528 int offset; 2529 2530 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2531 offset = (flag == SIT_BITMAP) ? 2532 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2533 /* 2534 * if large_nat_bitmap feature is enabled, leave checksum 2535 * protection for all nat/sit bitmaps. 2536 */ 2537 return tmp_ptr + offset + sizeof(__le32); 2538 } 2539 2540 if (__cp_payload(sbi) > 0) { 2541 if (flag == NAT_BITMAP) 2542 return tmp_ptr; 2543 else 2544 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2545 } else { 2546 offset = (flag == NAT_BITMAP) ? 2547 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2548 return tmp_ptr + offset; 2549 } 2550 } 2551 2552 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2553 { 2554 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2555 2556 if (sbi->cur_cp_pack == 2) 2557 start_addr += sbi->blocks_per_seg; 2558 return start_addr; 2559 } 2560 2561 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2562 { 2563 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2564 2565 if (sbi->cur_cp_pack == 1) 2566 start_addr += sbi->blocks_per_seg; 2567 return start_addr; 2568 } 2569 2570 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2571 { 2572 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2573 } 2574 2575 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2576 { 2577 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2578 } 2579 2580 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2581 struct inode *inode, bool is_inode) 2582 { 2583 block_t valid_block_count; 2584 unsigned int valid_node_count, user_block_count; 2585 int err; 2586 2587 if (is_inode) { 2588 if (inode) { 2589 err = dquot_alloc_inode(inode); 2590 if (err) 2591 return err; 2592 } 2593 } else { 2594 err = dquot_reserve_block(inode, 1); 2595 if (err) 2596 return err; 2597 } 2598 2599 if (time_to_inject(sbi, FAULT_BLOCK)) { 2600 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2601 goto enospc; 2602 } 2603 2604 spin_lock(&sbi->stat_lock); 2605 2606 valid_block_count = sbi->total_valid_block_count + 2607 sbi->current_reserved_blocks + 1; 2608 2609 if (!__allow_reserved_blocks(sbi, inode, false)) 2610 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2611 2612 if (F2FS_IO_ALIGNED(sbi)) 2613 valid_block_count += sbi->blocks_per_seg * 2614 SM_I(sbi)->additional_reserved_segments; 2615 2616 user_block_count = sbi->user_block_count; 2617 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2618 user_block_count -= sbi->unusable_block_count; 2619 2620 if (unlikely(valid_block_count > user_block_count)) { 2621 spin_unlock(&sbi->stat_lock); 2622 goto enospc; 2623 } 2624 2625 valid_node_count = sbi->total_valid_node_count + 1; 2626 if (unlikely(valid_node_count > sbi->total_node_count)) { 2627 spin_unlock(&sbi->stat_lock); 2628 goto enospc; 2629 } 2630 2631 sbi->total_valid_node_count++; 2632 sbi->total_valid_block_count++; 2633 spin_unlock(&sbi->stat_lock); 2634 2635 if (inode) { 2636 if (is_inode) 2637 f2fs_mark_inode_dirty_sync(inode, true); 2638 else 2639 f2fs_i_blocks_write(inode, 1, true, true); 2640 } 2641 2642 percpu_counter_inc(&sbi->alloc_valid_block_count); 2643 return 0; 2644 2645 enospc: 2646 if (is_inode) { 2647 if (inode) 2648 dquot_free_inode(inode); 2649 } else { 2650 dquot_release_reservation_block(inode, 1); 2651 } 2652 return -ENOSPC; 2653 } 2654 2655 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2656 struct inode *inode, bool is_inode) 2657 { 2658 spin_lock(&sbi->stat_lock); 2659 2660 if (unlikely(!sbi->total_valid_block_count || 2661 !sbi->total_valid_node_count)) { 2662 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2663 sbi->total_valid_block_count, 2664 sbi->total_valid_node_count); 2665 set_sbi_flag(sbi, SBI_NEED_FSCK); 2666 } else { 2667 sbi->total_valid_block_count--; 2668 sbi->total_valid_node_count--; 2669 } 2670 2671 if (sbi->reserved_blocks && 2672 sbi->current_reserved_blocks < sbi->reserved_blocks) 2673 sbi->current_reserved_blocks++; 2674 2675 spin_unlock(&sbi->stat_lock); 2676 2677 if (is_inode) { 2678 dquot_free_inode(inode); 2679 } else { 2680 if (unlikely(inode->i_blocks == 0)) { 2681 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2682 inode->i_ino, 2683 (unsigned long long)inode->i_blocks); 2684 set_sbi_flag(sbi, SBI_NEED_FSCK); 2685 return; 2686 } 2687 f2fs_i_blocks_write(inode, 1, false, true); 2688 } 2689 } 2690 2691 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2692 { 2693 return sbi->total_valid_node_count; 2694 } 2695 2696 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2697 { 2698 percpu_counter_inc(&sbi->total_valid_inode_count); 2699 } 2700 2701 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2702 { 2703 percpu_counter_dec(&sbi->total_valid_inode_count); 2704 } 2705 2706 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2707 { 2708 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2709 } 2710 2711 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2712 pgoff_t index, bool for_write) 2713 { 2714 struct page *page; 2715 unsigned int flags; 2716 2717 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2718 if (!for_write) 2719 page = find_get_page_flags(mapping, index, 2720 FGP_LOCK | FGP_ACCESSED); 2721 else 2722 page = find_lock_page(mapping, index); 2723 if (page) 2724 return page; 2725 2726 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2727 f2fs_show_injection_info(F2FS_M_SB(mapping), 2728 FAULT_PAGE_ALLOC); 2729 return NULL; 2730 } 2731 } 2732 2733 if (!for_write) 2734 return grab_cache_page(mapping, index); 2735 2736 flags = memalloc_nofs_save(); 2737 page = grab_cache_page_write_begin(mapping, index); 2738 memalloc_nofs_restore(flags); 2739 2740 return page; 2741 } 2742 2743 static inline struct page *f2fs_pagecache_get_page( 2744 struct address_space *mapping, pgoff_t index, 2745 int fgp_flags, gfp_t gfp_mask) 2746 { 2747 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2748 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2749 return NULL; 2750 } 2751 2752 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2753 } 2754 2755 static inline void f2fs_put_page(struct page *page, int unlock) 2756 { 2757 if (!page) 2758 return; 2759 2760 if (unlock) { 2761 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2762 unlock_page(page); 2763 } 2764 put_page(page); 2765 } 2766 2767 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2768 { 2769 if (dn->node_page) 2770 f2fs_put_page(dn->node_page, 1); 2771 if (dn->inode_page && dn->node_page != dn->inode_page) 2772 f2fs_put_page(dn->inode_page, 0); 2773 dn->node_page = NULL; 2774 dn->inode_page = NULL; 2775 } 2776 2777 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2778 size_t size) 2779 { 2780 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2781 } 2782 2783 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2784 gfp_t flags) 2785 { 2786 void *entry; 2787 2788 entry = kmem_cache_alloc(cachep, flags); 2789 if (!entry) 2790 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2791 return entry; 2792 } 2793 2794 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2795 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2796 { 2797 if (nofail) 2798 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2799 2800 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2801 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2802 return NULL; 2803 } 2804 2805 return kmem_cache_alloc(cachep, flags); 2806 } 2807 2808 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2809 { 2810 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2811 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2812 get_pages(sbi, F2FS_WB_CP_DATA) || 2813 get_pages(sbi, F2FS_DIO_READ) || 2814 get_pages(sbi, F2FS_DIO_WRITE)) 2815 return true; 2816 2817 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2818 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2819 return true; 2820 2821 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2822 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2823 return true; 2824 return false; 2825 } 2826 2827 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2828 { 2829 if (sbi->gc_mode == GC_URGENT_HIGH) 2830 return true; 2831 2832 if (is_inflight_io(sbi, type)) 2833 return false; 2834 2835 if (sbi->gc_mode == GC_URGENT_MID) 2836 return true; 2837 2838 if (sbi->gc_mode == GC_URGENT_LOW && 2839 (type == DISCARD_TIME || type == GC_TIME)) 2840 return true; 2841 2842 return f2fs_time_over(sbi, type); 2843 } 2844 2845 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2846 unsigned long index, void *item) 2847 { 2848 while (radix_tree_insert(root, index, item)) 2849 cond_resched(); 2850 } 2851 2852 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2853 2854 static inline bool IS_INODE(struct page *page) 2855 { 2856 struct f2fs_node *p = F2FS_NODE(page); 2857 2858 return RAW_IS_INODE(p); 2859 } 2860 2861 static inline int offset_in_addr(struct f2fs_inode *i) 2862 { 2863 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2864 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2865 } 2866 2867 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2868 { 2869 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2870 } 2871 2872 static inline int f2fs_has_extra_attr(struct inode *inode); 2873 static inline block_t data_blkaddr(struct inode *inode, 2874 struct page *node_page, unsigned int offset) 2875 { 2876 struct f2fs_node *raw_node; 2877 __le32 *addr_array; 2878 int base = 0; 2879 bool is_inode = IS_INODE(node_page); 2880 2881 raw_node = F2FS_NODE(node_page); 2882 2883 if (is_inode) { 2884 if (!inode) 2885 /* from GC path only */ 2886 base = offset_in_addr(&raw_node->i); 2887 else if (f2fs_has_extra_attr(inode)) 2888 base = get_extra_isize(inode); 2889 } 2890 2891 addr_array = blkaddr_in_node(raw_node); 2892 return le32_to_cpu(addr_array[base + offset]); 2893 } 2894 2895 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2896 { 2897 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2898 } 2899 2900 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2901 { 2902 int mask; 2903 2904 addr += (nr >> 3); 2905 mask = 1 << (7 - (nr & 0x07)); 2906 return mask & *addr; 2907 } 2908 2909 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2910 { 2911 int mask; 2912 2913 addr += (nr >> 3); 2914 mask = 1 << (7 - (nr & 0x07)); 2915 *addr |= mask; 2916 } 2917 2918 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2919 { 2920 int mask; 2921 2922 addr += (nr >> 3); 2923 mask = 1 << (7 - (nr & 0x07)); 2924 *addr &= ~mask; 2925 } 2926 2927 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2928 { 2929 int mask; 2930 int ret; 2931 2932 addr += (nr >> 3); 2933 mask = 1 << (7 - (nr & 0x07)); 2934 ret = mask & *addr; 2935 *addr |= mask; 2936 return ret; 2937 } 2938 2939 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2940 { 2941 int mask; 2942 int ret; 2943 2944 addr += (nr >> 3); 2945 mask = 1 << (7 - (nr & 0x07)); 2946 ret = mask & *addr; 2947 *addr &= ~mask; 2948 return ret; 2949 } 2950 2951 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2952 { 2953 int mask; 2954 2955 addr += (nr >> 3); 2956 mask = 1 << (7 - (nr & 0x07)); 2957 *addr ^= mask; 2958 } 2959 2960 /* 2961 * On-disk inode flags (f2fs_inode::i_flags) 2962 */ 2963 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2964 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2965 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2966 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2967 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2968 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2969 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2970 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2971 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2972 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2973 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2974 2975 /* Flags that should be inherited by new inodes from their parent. */ 2976 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2977 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2978 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2979 2980 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2981 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2982 F2FS_CASEFOLD_FL)) 2983 2984 /* Flags that are appropriate for non-directories/regular files. */ 2985 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2986 2987 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2988 { 2989 if (S_ISDIR(mode)) 2990 return flags; 2991 else if (S_ISREG(mode)) 2992 return flags & F2FS_REG_FLMASK; 2993 else 2994 return flags & F2FS_OTHER_FLMASK; 2995 } 2996 2997 static inline void __mark_inode_dirty_flag(struct inode *inode, 2998 int flag, bool set) 2999 { 3000 switch (flag) { 3001 case FI_INLINE_XATTR: 3002 case FI_INLINE_DATA: 3003 case FI_INLINE_DENTRY: 3004 case FI_NEW_INODE: 3005 if (set) 3006 return; 3007 fallthrough; 3008 case FI_DATA_EXIST: 3009 case FI_INLINE_DOTS: 3010 case FI_PIN_FILE: 3011 case FI_COMPRESS_RELEASED: 3012 f2fs_mark_inode_dirty_sync(inode, true); 3013 } 3014 } 3015 3016 static inline void set_inode_flag(struct inode *inode, int flag) 3017 { 3018 set_bit(flag, F2FS_I(inode)->flags); 3019 __mark_inode_dirty_flag(inode, flag, true); 3020 } 3021 3022 static inline int is_inode_flag_set(struct inode *inode, int flag) 3023 { 3024 return test_bit(flag, F2FS_I(inode)->flags); 3025 } 3026 3027 static inline void clear_inode_flag(struct inode *inode, int flag) 3028 { 3029 clear_bit(flag, F2FS_I(inode)->flags); 3030 __mark_inode_dirty_flag(inode, flag, false); 3031 } 3032 3033 static inline bool f2fs_verity_in_progress(struct inode *inode) 3034 { 3035 return IS_ENABLED(CONFIG_FS_VERITY) && 3036 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3037 } 3038 3039 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3040 { 3041 F2FS_I(inode)->i_acl_mode = mode; 3042 set_inode_flag(inode, FI_ACL_MODE); 3043 f2fs_mark_inode_dirty_sync(inode, false); 3044 } 3045 3046 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3047 { 3048 if (inc) 3049 inc_nlink(inode); 3050 else 3051 drop_nlink(inode); 3052 f2fs_mark_inode_dirty_sync(inode, true); 3053 } 3054 3055 static inline void f2fs_i_blocks_write(struct inode *inode, 3056 block_t diff, bool add, bool claim) 3057 { 3058 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3059 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3060 3061 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3062 if (add) { 3063 if (claim) 3064 dquot_claim_block(inode, diff); 3065 else 3066 dquot_alloc_block_nofail(inode, diff); 3067 } else { 3068 dquot_free_block(inode, diff); 3069 } 3070 3071 f2fs_mark_inode_dirty_sync(inode, true); 3072 if (clean || recover) 3073 set_inode_flag(inode, FI_AUTO_RECOVER); 3074 } 3075 3076 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3077 { 3078 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3079 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3080 3081 if (i_size_read(inode) == i_size) 3082 return; 3083 3084 i_size_write(inode, i_size); 3085 f2fs_mark_inode_dirty_sync(inode, true); 3086 if (clean || recover) 3087 set_inode_flag(inode, FI_AUTO_RECOVER); 3088 } 3089 3090 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3091 { 3092 F2FS_I(inode)->i_current_depth = depth; 3093 f2fs_mark_inode_dirty_sync(inode, true); 3094 } 3095 3096 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3097 unsigned int count) 3098 { 3099 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3100 f2fs_mark_inode_dirty_sync(inode, true); 3101 } 3102 3103 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3104 { 3105 F2FS_I(inode)->i_xattr_nid = xnid; 3106 f2fs_mark_inode_dirty_sync(inode, true); 3107 } 3108 3109 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3110 { 3111 F2FS_I(inode)->i_pino = pino; 3112 f2fs_mark_inode_dirty_sync(inode, true); 3113 } 3114 3115 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3116 { 3117 struct f2fs_inode_info *fi = F2FS_I(inode); 3118 3119 if (ri->i_inline & F2FS_INLINE_XATTR) 3120 set_bit(FI_INLINE_XATTR, fi->flags); 3121 if (ri->i_inline & F2FS_INLINE_DATA) 3122 set_bit(FI_INLINE_DATA, fi->flags); 3123 if (ri->i_inline & F2FS_INLINE_DENTRY) 3124 set_bit(FI_INLINE_DENTRY, fi->flags); 3125 if (ri->i_inline & F2FS_DATA_EXIST) 3126 set_bit(FI_DATA_EXIST, fi->flags); 3127 if (ri->i_inline & F2FS_INLINE_DOTS) 3128 set_bit(FI_INLINE_DOTS, fi->flags); 3129 if (ri->i_inline & F2FS_EXTRA_ATTR) 3130 set_bit(FI_EXTRA_ATTR, fi->flags); 3131 if (ri->i_inline & F2FS_PIN_FILE) 3132 set_bit(FI_PIN_FILE, fi->flags); 3133 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3134 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3135 } 3136 3137 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3138 { 3139 ri->i_inline = 0; 3140 3141 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3142 ri->i_inline |= F2FS_INLINE_XATTR; 3143 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3144 ri->i_inline |= F2FS_INLINE_DATA; 3145 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3146 ri->i_inline |= F2FS_INLINE_DENTRY; 3147 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3148 ri->i_inline |= F2FS_DATA_EXIST; 3149 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3150 ri->i_inline |= F2FS_INLINE_DOTS; 3151 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3152 ri->i_inline |= F2FS_EXTRA_ATTR; 3153 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3154 ri->i_inline |= F2FS_PIN_FILE; 3155 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3156 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3157 } 3158 3159 static inline int f2fs_has_extra_attr(struct inode *inode) 3160 { 3161 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3162 } 3163 3164 static inline int f2fs_has_inline_xattr(struct inode *inode) 3165 { 3166 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3167 } 3168 3169 static inline int f2fs_compressed_file(struct inode *inode) 3170 { 3171 return S_ISREG(inode->i_mode) && 3172 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3173 } 3174 3175 static inline bool f2fs_need_compress_data(struct inode *inode) 3176 { 3177 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3178 3179 if (!f2fs_compressed_file(inode)) 3180 return false; 3181 3182 if (compress_mode == COMPR_MODE_FS) 3183 return true; 3184 else if (compress_mode == COMPR_MODE_USER && 3185 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3186 return true; 3187 3188 return false; 3189 } 3190 3191 static inline unsigned int addrs_per_inode(struct inode *inode) 3192 { 3193 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3194 get_inline_xattr_addrs(inode); 3195 3196 if (!f2fs_compressed_file(inode)) 3197 return addrs; 3198 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3199 } 3200 3201 static inline unsigned int addrs_per_block(struct inode *inode) 3202 { 3203 if (!f2fs_compressed_file(inode)) 3204 return DEF_ADDRS_PER_BLOCK; 3205 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3206 } 3207 3208 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3209 { 3210 struct f2fs_inode *ri = F2FS_INODE(page); 3211 3212 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3213 get_inline_xattr_addrs(inode)]); 3214 } 3215 3216 static inline int inline_xattr_size(struct inode *inode) 3217 { 3218 if (f2fs_has_inline_xattr(inode)) 3219 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3220 return 0; 3221 } 3222 3223 /* 3224 * Notice: check inline_data flag without inode page lock is unsafe. 3225 * It could change at any time by f2fs_convert_inline_page(). 3226 */ 3227 static inline int f2fs_has_inline_data(struct inode *inode) 3228 { 3229 return is_inode_flag_set(inode, FI_INLINE_DATA); 3230 } 3231 3232 static inline int f2fs_exist_data(struct inode *inode) 3233 { 3234 return is_inode_flag_set(inode, FI_DATA_EXIST); 3235 } 3236 3237 static inline int f2fs_has_inline_dots(struct inode *inode) 3238 { 3239 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3240 } 3241 3242 static inline int f2fs_is_mmap_file(struct inode *inode) 3243 { 3244 return is_inode_flag_set(inode, FI_MMAP_FILE); 3245 } 3246 3247 static inline bool f2fs_is_pinned_file(struct inode *inode) 3248 { 3249 return is_inode_flag_set(inode, FI_PIN_FILE); 3250 } 3251 3252 static inline bool f2fs_is_atomic_file(struct inode *inode) 3253 { 3254 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3255 } 3256 3257 static inline bool f2fs_is_cow_file(struct inode *inode) 3258 { 3259 return is_inode_flag_set(inode, FI_COW_FILE); 3260 } 3261 3262 static inline bool f2fs_is_first_block_written(struct inode *inode) 3263 { 3264 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3265 } 3266 3267 static inline bool f2fs_is_drop_cache(struct inode *inode) 3268 { 3269 return is_inode_flag_set(inode, FI_DROP_CACHE); 3270 } 3271 3272 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3273 { 3274 struct f2fs_inode *ri = F2FS_INODE(page); 3275 int extra_size = get_extra_isize(inode); 3276 3277 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3278 } 3279 3280 static inline int f2fs_has_inline_dentry(struct inode *inode) 3281 { 3282 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3283 } 3284 3285 static inline int is_file(struct inode *inode, int type) 3286 { 3287 return F2FS_I(inode)->i_advise & type; 3288 } 3289 3290 static inline void set_file(struct inode *inode, int type) 3291 { 3292 if (is_file(inode, type)) 3293 return; 3294 F2FS_I(inode)->i_advise |= type; 3295 f2fs_mark_inode_dirty_sync(inode, true); 3296 } 3297 3298 static inline void clear_file(struct inode *inode, int type) 3299 { 3300 if (!is_file(inode, type)) 3301 return; 3302 F2FS_I(inode)->i_advise &= ~type; 3303 f2fs_mark_inode_dirty_sync(inode, true); 3304 } 3305 3306 static inline bool f2fs_is_time_consistent(struct inode *inode) 3307 { 3308 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3309 return false; 3310 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3311 return false; 3312 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3313 return false; 3314 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3315 &F2FS_I(inode)->i_crtime)) 3316 return false; 3317 return true; 3318 } 3319 3320 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3321 { 3322 bool ret; 3323 3324 if (dsync) { 3325 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3326 3327 spin_lock(&sbi->inode_lock[DIRTY_META]); 3328 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3329 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3330 return ret; 3331 } 3332 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3333 file_keep_isize(inode) || 3334 i_size_read(inode) & ~PAGE_MASK) 3335 return false; 3336 3337 if (!f2fs_is_time_consistent(inode)) 3338 return false; 3339 3340 spin_lock(&F2FS_I(inode)->i_size_lock); 3341 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3342 spin_unlock(&F2FS_I(inode)->i_size_lock); 3343 3344 return ret; 3345 } 3346 3347 static inline bool f2fs_readonly(struct super_block *sb) 3348 { 3349 return sb_rdonly(sb); 3350 } 3351 3352 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3353 { 3354 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3355 } 3356 3357 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3358 { 3359 if (len == 1 && name[0] == '.') 3360 return true; 3361 3362 if (len == 2 && name[0] == '.' && name[1] == '.') 3363 return true; 3364 3365 return false; 3366 } 3367 3368 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3369 size_t size, gfp_t flags) 3370 { 3371 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3372 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3373 return NULL; 3374 } 3375 3376 return kmalloc(size, flags); 3377 } 3378 3379 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3380 size_t size, gfp_t flags) 3381 { 3382 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3383 } 3384 3385 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3386 size_t size, gfp_t flags) 3387 { 3388 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3389 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3390 return NULL; 3391 } 3392 3393 return kvmalloc(size, flags); 3394 } 3395 3396 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3397 size_t size, gfp_t flags) 3398 { 3399 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3400 } 3401 3402 static inline int get_extra_isize(struct inode *inode) 3403 { 3404 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3405 } 3406 3407 static inline int get_inline_xattr_addrs(struct inode *inode) 3408 { 3409 return F2FS_I(inode)->i_inline_xattr_size; 3410 } 3411 3412 #define f2fs_get_inode_mode(i) \ 3413 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3414 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3415 3416 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3417 (offsetof(struct f2fs_inode, i_extra_end) - \ 3418 offsetof(struct f2fs_inode, i_extra_isize)) \ 3419 3420 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3421 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3422 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3423 sizeof((f2fs_inode)->field)) \ 3424 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3425 3426 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3427 3428 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3429 3430 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3431 block_t blkaddr, int type); 3432 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3433 block_t blkaddr, int type) 3434 { 3435 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3436 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3437 blkaddr, type); 3438 f2fs_bug_on(sbi, 1); 3439 } 3440 } 3441 3442 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3443 { 3444 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3445 blkaddr == COMPRESS_ADDR) 3446 return false; 3447 return true; 3448 } 3449 3450 /* 3451 * file.c 3452 */ 3453 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3454 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3455 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3456 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3457 int f2fs_truncate(struct inode *inode); 3458 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3459 struct kstat *stat, u32 request_mask, unsigned int flags); 3460 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3461 struct iattr *attr); 3462 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3463 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3464 int f2fs_precache_extents(struct inode *inode); 3465 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3466 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3467 struct dentry *dentry, struct fileattr *fa); 3468 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3469 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3470 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3471 int f2fs_pin_file_control(struct inode *inode, bool inc); 3472 3473 /* 3474 * inode.c 3475 */ 3476 void f2fs_set_inode_flags(struct inode *inode); 3477 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3478 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3479 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3480 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3481 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3482 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3483 void f2fs_update_inode_page(struct inode *inode); 3484 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3485 void f2fs_evict_inode(struct inode *inode); 3486 void f2fs_handle_failed_inode(struct inode *inode); 3487 3488 /* 3489 * namei.c 3490 */ 3491 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3492 bool hot, bool set); 3493 struct dentry *f2fs_get_parent(struct dentry *child); 3494 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3495 struct inode **new_inode); 3496 3497 /* 3498 * dir.c 3499 */ 3500 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3501 int f2fs_init_casefolded_name(const struct inode *dir, 3502 struct f2fs_filename *fname); 3503 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3504 int lookup, struct f2fs_filename *fname); 3505 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3506 struct f2fs_filename *fname); 3507 void f2fs_free_filename(struct f2fs_filename *fname); 3508 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3509 const struct f2fs_filename *fname, int *max_slots); 3510 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3511 unsigned int start_pos, struct fscrypt_str *fstr); 3512 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3513 struct f2fs_dentry_ptr *d); 3514 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3515 const struct f2fs_filename *fname, struct page *dpage); 3516 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3517 unsigned int current_depth); 3518 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3519 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3520 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3521 const struct f2fs_filename *fname, 3522 struct page **res_page); 3523 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3524 const struct qstr *child, struct page **res_page); 3525 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3526 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3527 struct page **page); 3528 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3529 struct page *page, struct inode *inode); 3530 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3531 const struct f2fs_filename *fname); 3532 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3533 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3534 unsigned int bit_pos); 3535 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3536 struct inode *inode, nid_t ino, umode_t mode); 3537 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3538 struct inode *inode, nid_t ino, umode_t mode); 3539 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3540 struct inode *inode, nid_t ino, umode_t mode); 3541 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3542 struct inode *dir, struct inode *inode); 3543 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3544 bool f2fs_empty_dir(struct inode *dir); 3545 3546 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3547 { 3548 if (fscrypt_is_nokey_name(dentry)) 3549 return -ENOKEY; 3550 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3551 inode, inode->i_ino, inode->i_mode); 3552 } 3553 3554 /* 3555 * super.c 3556 */ 3557 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3558 void f2fs_inode_synced(struct inode *inode); 3559 int f2fs_dquot_initialize(struct inode *inode); 3560 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3561 int f2fs_quota_sync(struct super_block *sb, int type); 3562 loff_t max_file_blocks(struct inode *inode); 3563 void f2fs_quota_off_umount(struct super_block *sb); 3564 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason); 3565 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3566 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3567 int f2fs_sync_fs(struct super_block *sb, int sync); 3568 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3569 3570 /* 3571 * hash.c 3572 */ 3573 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3574 3575 /* 3576 * node.c 3577 */ 3578 struct node_info; 3579 3580 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3581 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3582 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3583 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3584 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3585 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3586 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3587 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3588 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3589 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3590 struct node_info *ni, bool checkpoint_context); 3591 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3592 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3593 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3594 int f2fs_truncate_xattr_node(struct inode *inode); 3595 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3596 unsigned int seq_id); 3597 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3598 int f2fs_remove_inode_page(struct inode *inode); 3599 struct page *f2fs_new_inode_page(struct inode *inode); 3600 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3601 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3602 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3603 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3604 int f2fs_move_node_page(struct page *node_page, int gc_type); 3605 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3606 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3607 struct writeback_control *wbc, bool atomic, 3608 unsigned int *seq_id); 3609 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3610 struct writeback_control *wbc, 3611 bool do_balance, enum iostat_type io_type); 3612 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3613 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3614 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3615 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3616 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3617 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3618 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3619 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3620 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3621 unsigned int segno, struct f2fs_summary_block *sum); 3622 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3623 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3624 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3625 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3626 int __init f2fs_create_node_manager_caches(void); 3627 void f2fs_destroy_node_manager_caches(void); 3628 3629 /* 3630 * segment.c 3631 */ 3632 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3633 int f2fs_commit_atomic_write(struct inode *inode); 3634 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3635 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3636 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3637 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3638 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3639 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3640 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3641 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3642 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3643 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3644 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3645 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3646 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3647 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3648 struct cp_control *cpc); 3649 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3650 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3651 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3652 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3653 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3654 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3655 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3656 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3657 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3658 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3659 unsigned int *newseg, bool new_sec, int dir); 3660 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3661 unsigned int start, unsigned int end); 3662 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3663 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3664 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3665 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3666 struct cp_control *cpc); 3667 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3668 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3669 block_t blk_addr); 3670 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3671 enum iostat_type io_type); 3672 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3673 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3674 struct f2fs_io_info *fio); 3675 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3676 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3677 block_t old_blkaddr, block_t new_blkaddr, 3678 bool recover_curseg, bool recover_newaddr, 3679 bool from_gc); 3680 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3681 block_t old_addr, block_t new_addr, 3682 unsigned char version, bool recover_curseg, 3683 bool recover_newaddr); 3684 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3685 block_t old_blkaddr, block_t *new_blkaddr, 3686 struct f2fs_summary *sum, int type, 3687 struct f2fs_io_info *fio); 3688 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3689 block_t blkaddr, unsigned int blkcnt); 3690 void f2fs_wait_on_page_writeback(struct page *page, 3691 enum page_type type, bool ordered, bool locked); 3692 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3693 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3694 block_t len); 3695 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3696 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3697 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3698 unsigned int val, int alloc); 3699 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3700 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3701 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3702 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3703 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3704 int __init f2fs_create_segment_manager_caches(void); 3705 void f2fs_destroy_segment_manager_caches(void); 3706 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3707 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3708 unsigned int segno); 3709 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3710 unsigned int segno); 3711 3712 #define DEF_FRAGMENT_SIZE 4 3713 #define MIN_FRAGMENT_SIZE 1 3714 #define MAX_FRAGMENT_SIZE 512 3715 3716 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3717 { 3718 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3719 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3720 } 3721 3722 /* 3723 * checkpoint.c 3724 */ 3725 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3726 unsigned char reason); 3727 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3728 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3729 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3730 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3731 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3732 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3733 block_t blkaddr, int type); 3734 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3735 int type, bool sync); 3736 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3737 unsigned int ra_blocks); 3738 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3739 long nr_to_write, enum iostat_type io_type); 3740 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3741 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3742 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3743 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3744 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3745 unsigned int devidx, int type); 3746 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3747 unsigned int devidx, int type); 3748 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3749 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3750 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3751 void f2fs_add_orphan_inode(struct inode *inode); 3752 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3753 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3754 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3755 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3756 void f2fs_remove_dirty_inode(struct inode *inode); 3757 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3758 bool from_cp); 3759 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3760 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3761 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3762 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3763 int __init f2fs_create_checkpoint_caches(void); 3764 void f2fs_destroy_checkpoint_caches(void); 3765 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3766 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3767 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3768 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3769 3770 /* 3771 * data.c 3772 */ 3773 int __init f2fs_init_bioset(void); 3774 void f2fs_destroy_bioset(void); 3775 int f2fs_init_bio_entry_cache(void); 3776 void f2fs_destroy_bio_entry_cache(void); 3777 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3778 struct bio *bio, enum page_type type); 3779 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3780 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3781 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3782 struct inode *inode, struct page *page, 3783 nid_t ino, enum page_type type); 3784 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3785 struct bio **bio, struct page *page); 3786 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3787 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3788 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3789 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3790 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3791 block_t blk_addr, sector_t *sector); 3792 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3793 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3794 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3795 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3796 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3797 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3798 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3799 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3800 blk_opf_t op_flags, bool for_write); 3801 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3802 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3803 bool for_write); 3804 struct page *f2fs_get_new_data_page(struct inode *inode, 3805 struct page *ipage, pgoff_t index, bool new_i_size); 3806 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3807 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3808 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3809 int create, int flag); 3810 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3811 u64 start, u64 len); 3812 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3813 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3814 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3815 int f2fs_write_single_data_page(struct page *page, int *submitted, 3816 struct bio **bio, sector_t *last_block, 3817 struct writeback_control *wbc, 3818 enum iostat_type io_type, 3819 int compr_blocks, bool allow_balance); 3820 void f2fs_write_failed(struct inode *inode, loff_t to); 3821 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3822 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3823 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3824 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3825 int f2fs_init_post_read_processing(void); 3826 void f2fs_destroy_post_read_processing(void); 3827 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3828 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3829 extern const struct iomap_ops f2fs_iomap_ops; 3830 3831 /* 3832 * gc.c 3833 */ 3834 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3835 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3836 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3837 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3838 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3839 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3840 int __init f2fs_create_garbage_collection_cache(void); 3841 void f2fs_destroy_garbage_collection_cache(void); 3842 3843 /* 3844 * recovery.c 3845 */ 3846 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3847 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3848 int __init f2fs_create_recovery_cache(void); 3849 void f2fs_destroy_recovery_cache(void); 3850 3851 /* 3852 * debug.c 3853 */ 3854 #ifdef CONFIG_F2FS_STAT_FS 3855 struct f2fs_stat_info { 3856 struct list_head stat_list; 3857 struct f2fs_sb_info *sbi; 3858 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3859 int main_area_segs, main_area_sections, main_area_zones; 3860 unsigned long long hit_largest, hit_cached, hit_rbtree; 3861 unsigned long long hit_total, total_ext; 3862 int ext_tree, zombie_tree, ext_node; 3863 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3864 int ndirty_data, ndirty_qdata; 3865 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3866 int nats, dirty_nats, sits, dirty_sits; 3867 int free_nids, avail_nids, alloc_nids; 3868 int total_count, utilization; 3869 int bg_gc, nr_wb_cp_data, nr_wb_data; 3870 int nr_rd_data, nr_rd_node, nr_rd_meta; 3871 int nr_dio_read, nr_dio_write; 3872 unsigned int io_skip_bggc, other_skip_bggc; 3873 int nr_flushing, nr_flushed, flush_list_empty; 3874 int nr_discarding, nr_discarded; 3875 int nr_discard_cmd; 3876 unsigned int undiscard_blks; 3877 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3878 unsigned int cur_ckpt_time, peak_ckpt_time; 3879 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3880 int compr_inode, swapfile_inode; 3881 unsigned long long compr_blocks; 3882 int aw_cnt, max_aw_cnt; 3883 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3884 unsigned int bimodal, avg_vblocks; 3885 int util_free, util_valid, util_invalid; 3886 int rsvd_segs, overp_segs; 3887 int dirty_count, node_pages, meta_pages, compress_pages; 3888 int compress_page_hit; 3889 int prefree_count, call_count, cp_count, bg_cp_count; 3890 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3891 int bg_node_segs, bg_data_segs; 3892 int tot_blks, data_blks, node_blks; 3893 int bg_data_blks, bg_node_blks; 3894 int curseg[NR_CURSEG_TYPE]; 3895 int cursec[NR_CURSEG_TYPE]; 3896 int curzone[NR_CURSEG_TYPE]; 3897 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3898 unsigned int full_seg[NR_CURSEG_TYPE]; 3899 unsigned int valid_blks[NR_CURSEG_TYPE]; 3900 3901 unsigned int meta_count[META_MAX]; 3902 unsigned int segment_count[2]; 3903 unsigned int block_count[2]; 3904 unsigned int inplace_count; 3905 unsigned long long base_mem, cache_mem, page_mem; 3906 }; 3907 3908 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3909 { 3910 return (struct f2fs_stat_info *)sbi->stat_info; 3911 } 3912 3913 #define stat_inc_cp_count(si) ((si)->cp_count++) 3914 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3915 #define stat_inc_call_count(si) ((si)->call_count++) 3916 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3917 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3918 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3919 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3920 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3921 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3922 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3923 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3924 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3925 #define stat_inc_inline_xattr(inode) \ 3926 do { \ 3927 if (f2fs_has_inline_xattr(inode)) \ 3928 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3929 } while (0) 3930 #define stat_dec_inline_xattr(inode) \ 3931 do { \ 3932 if (f2fs_has_inline_xattr(inode)) \ 3933 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3934 } while (0) 3935 #define stat_inc_inline_inode(inode) \ 3936 do { \ 3937 if (f2fs_has_inline_data(inode)) \ 3938 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3939 } while (0) 3940 #define stat_dec_inline_inode(inode) \ 3941 do { \ 3942 if (f2fs_has_inline_data(inode)) \ 3943 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3944 } while (0) 3945 #define stat_inc_inline_dir(inode) \ 3946 do { \ 3947 if (f2fs_has_inline_dentry(inode)) \ 3948 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3949 } while (0) 3950 #define stat_dec_inline_dir(inode) \ 3951 do { \ 3952 if (f2fs_has_inline_dentry(inode)) \ 3953 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3954 } while (0) 3955 #define stat_inc_compr_inode(inode) \ 3956 do { \ 3957 if (f2fs_compressed_file(inode)) \ 3958 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3959 } while (0) 3960 #define stat_dec_compr_inode(inode) \ 3961 do { \ 3962 if (f2fs_compressed_file(inode)) \ 3963 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3964 } while (0) 3965 #define stat_add_compr_blocks(inode, blocks) \ 3966 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3967 #define stat_sub_compr_blocks(inode, blocks) \ 3968 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3969 #define stat_inc_swapfile_inode(inode) \ 3970 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 3971 #define stat_dec_swapfile_inode(inode) \ 3972 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 3973 #define stat_inc_atomic_inode(inode) \ 3974 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 3975 #define stat_dec_atomic_inode(inode) \ 3976 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 3977 #define stat_inc_meta_count(sbi, blkaddr) \ 3978 do { \ 3979 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3980 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3981 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3982 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3983 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3984 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3985 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3986 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3987 } while (0) 3988 #define stat_inc_seg_type(sbi, curseg) \ 3989 ((sbi)->segment_count[(curseg)->alloc_type]++) 3990 #define stat_inc_block_count(sbi, curseg) \ 3991 ((sbi)->block_count[(curseg)->alloc_type]++) 3992 #define stat_inc_inplace_blocks(sbi) \ 3993 (atomic_inc(&(sbi)->inplace_count)) 3994 #define stat_update_max_atomic_write(inode) \ 3995 do { \ 3996 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 3997 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3998 if (cur > max) \ 3999 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4000 } while (0) 4001 #define stat_inc_seg_count(sbi, type, gc_type) \ 4002 do { \ 4003 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4004 si->tot_segs++; \ 4005 if ((type) == SUM_TYPE_DATA) { \ 4006 si->data_segs++; \ 4007 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 4008 } else { \ 4009 si->node_segs++; \ 4010 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 4011 } \ 4012 } while (0) 4013 4014 #define stat_inc_tot_blk_count(si, blks) \ 4015 ((si)->tot_blks += (blks)) 4016 4017 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4018 do { \ 4019 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4020 stat_inc_tot_blk_count(si, blks); \ 4021 si->data_blks += (blks); \ 4022 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4023 } while (0) 4024 4025 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4026 do { \ 4027 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4028 stat_inc_tot_blk_count(si, blks); \ 4029 si->node_blks += (blks); \ 4030 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4031 } while (0) 4032 4033 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4034 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4035 void __init f2fs_create_root_stats(void); 4036 void f2fs_destroy_root_stats(void); 4037 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4038 #else 4039 #define stat_inc_cp_count(si) do { } while (0) 4040 #define stat_inc_bg_cp_count(si) do { } while (0) 4041 #define stat_inc_call_count(si) do { } while (0) 4042 #define stat_inc_bggc_count(si) do { } while (0) 4043 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4044 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4045 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4046 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4047 #define stat_inc_total_hit(sbi) do { } while (0) 4048 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 4049 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4050 #define stat_inc_cached_node_hit(sbi) do { } while (0) 4051 #define stat_inc_inline_xattr(inode) do { } while (0) 4052 #define stat_dec_inline_xattr(inode) do { } while (0) 4053 #define stat_inc_inline_inode(inode) do { } while (0) 4054 #define stat_dec_inline_inode(inode) do { } while (0) 4055 #define stat_inc_inline_dir(inode) do { } while (0) 4056 #define stat_dec_inline_dir(inode) do { } while (0) 4057 #define stat_inc_compr_inode(inode) do { } while (0) 4058 #define stat_dec_compr_inode(inode) do { } while (0) 4059 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4060 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4061 #define stat_inc_swapfile_inode(inode) do { } while (0) 4062 #define stat_dec_swapfile_inode(inode) do { } while (0) 4063 #define stat_inc_atomic_inode(inode) do { } while (0) 4064 #define stat_dec_atomic_inode(inode) do { } while (0) 4065 #define stat_update_max_atomic_write(inode) do { } while (0) 4066 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4067 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4068 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4069 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4070 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4071 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4072 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4073 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4074 4075 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4076 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4077 static inline void __init f2fs_create_root_stats(void) { } 4078 static inline void f2fs_destroy_root_stats(void) { } 4079 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4080 #endif 4081 4082 extern const struct file_operations f2fs_dir_operations; 4083 extern const struct file_operations f2fs_file_operations; 4084 extern const struct inode_operations f2fs_file_inode_operations; 4085 extern const struct address_space_operations f2fs_dblock_aops; 4086 extern const struct address_space_operations f2fs_node_aops; 4087 extern const struct address_space_operations f2fs_meta_aops; 4088 extern const struct inode_operations f2fs_dir_inode_operations; 4089 extern const struct inode_operations f2fs_symlink_inode_operations; 4090 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4091 extern const struct inode_operations f2fs_special_inode_operations; 4092 extern struct kmem_cache *f2fs_inode_entry_slab; 4093 4094 /* 4095 * inline.c 4096 */ 4097 bool f2fs_may_inline_data(struct inode *inode); 4098 bool f2fs_sanity_check_inline_data(struct inode *inode); 4099 bool f2fs_may_inline_dentry(struct inode *inode); 4100 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4101 void f2fs_truncate_inline_inode(struct inode *inode, 4102 struct page *ipage, u64 from); 4103 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4104 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4105 int f2fs_convert_inline_inode(struct inode *inode); 4106 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4107 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4108 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4109 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4110 const struct f2fs_filename *fname, 4111 struct page **res_page); 4112 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4113 struct page *ipage); 4114 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4115 struct inode *inode, nid_t ino, umode_t mode); 4116 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4117 struct page *page, struct inode *dir, 4118 struct inode *inode); 4119 bool f2fs_empty_inline_dir(struct inode *dir); 4120 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4121 struct fscrypt_str *fstr); 4122 int f2fs_inline_data_fiemap(struct inode *inode, 4123 struct fiemap_extent_info *fieinfo, 4124 __u64 start, __u64 len); 4125 4126 /* 4127 * shrinker.c 4128 */ 4129 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4130 struct shrink_control *sc); 4131 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4132 struct shrink_control *sc); 4133 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4134 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4135 4136 /* 4137 * extent_cache.c 4138 */ 4139 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4140 struct rb_entry *cached_re, unsigned int ofs); 4141 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4142 struct rb_root_cached *root, 4143 struct rb_node **parent, 4144 unsigned long long key, bool *left_most); 4145 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4146 struct rb_root_cached *root, 4147 struct rb_node **parent, 4148 unsigned int ofs, bool *leftmost); 4149 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4150 struct rb_entry *cached_re, unsigned int ofs, 4151 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4152 struct rb_node ***insert_p, struct rb_node **insert_parent, 4153 bool force, bool *leftmost); 4154 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4155 struct rb_root_cached *root, bool check_key); 4156 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4157 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4158 void f2fs_drop_extent_tree(struct inode *inode); 4159 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4160 void f2fs_destroy_extent_tree(struct inode *inode); 4161 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4162 struct extent_info *ei); 4163 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4164 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4165 pgoff_t fofs, block_t blkaddr, unsigned int len); 4166 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4167 int __init f2fs_create_extent_cache(void); 4168 void f2fs_destroy_extent_cache(void); 4169 4170 /* 4171 * sysfs.c 4172 */ 4173 #define MIN_RA_MUL 2 4174 #define MAX_RA_MUL 256 4175 4176 int __init f2fs_init_sysfs(void); 4177 void f2fs_exit_sysfs(void); 4178 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4179 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4180 4181 /* verity.c */ 4182 extern const struct fsverity_operations f2fs_verityops; 4183 4184 /* 4185 * crypto support 4186 */ 4187 static inline bool f2fs_encrypted_file(struct inode *inode) 4188 { 4189 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4190 } 4191 4192 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4193 { 4194 #ifdef CONFIG_FS_ENCRYPTION 4195 file_set_encrypt(inode); 4196 f2fs_set_inode_flags(inode); 4197 #endif 4198 } 4199 4200 /* 4201 * Returns true if the reads of the inode's data need to undergo some 4202 * postprocessing step, like decryption or authenticity verification. 4203 */ 4204 static inline bool f2fs_post_read_required(struct inode *inode) 4205 { 4206 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4207 f2fs_compressed_file(inode); 4208 } 4209 4210 /* 4211 * compress.c 4212 */ 4213 #ifdef CONFIG_F2FS_FS_COMPRESSION 4214 bool f2fs_is_compressed_page(struct page *page); 4215 struct page *f2fs_compress_control_page(struct page *page); 4216 int f2fs_prepare_compress_overwrite(struct inode *inode, 4217 struct page **pagep, pgoff_t index, void **fsdata); 4218 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4219 pgoff_t index, unsigned copied); 4220 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4221 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4222 bool f2fs_is_compress_backend_ready(struct inode *inode); 4223 int f2fs_init_compress_mempool(void); 4224 void f2fs_destroy_compress_mempool(void); 4225 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4226 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4227 block_t blkaddr, bool in_task); 4228 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4229 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4230 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4231 int index, int nr_pages, bool uptodate); 4232 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4233 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4234 int f2fs_write_multi_pages(struct compress_ctx *cc, 4235 int *submitted, 4236 struct writeback_control *wbc, 4237 enum iostat_type io_type); 4238 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4239 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4240 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4241 unsigned int c_len); 4242 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4243 unsigned nr_pages, sector_t *last_block_in_bio, 4244 bool is_readahead, bool for_write); 4245 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4246 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4247 bool in_task); 4248 void f2fs_put_page_dic(struct page *page, bool in_task); 4249 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4250 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4251 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4252 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4253 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4254 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4255 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4256 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4257 int __init f2fs_init_compress_cache(void); 4258 void f2fs_destroy_compress_cache(void); 4259 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4260 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4261 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4262 nid_t ino, block_t blkaddr); 4263 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4264 block_t blkaddr); 4265 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4266 #define inc_compr_inode_stat(inode) \ 4267 do { \ 4268 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4269 sbi->compr_new_inode++; \ 4270 } while (0) 4271 #define add_compr_block_stat(inode, blocks) \ 4272 do { \ 4273 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4274 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4275 sbi->compr_written_block += blocks; \ 4276 sbi->compr_saved_block += diff; \ 4277 } while (0) 4278 #else 4279 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4280 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4281 { 4282 if (!f2fs_compressed_file(inode)) 4283 return true; 4284 /* not support compression */ 4285 return false; 4286 } 4287 static inline struct page *f2fs_compress_control_page(struct page *page) 4288 { 4289 WARN_ON_ONCE(1); 4290 return ERR_PTR(-EINVAL); 4291 } 4292 static inline int f2fs_init_compress_mempool(void) { return 0; } 4293 static inline void f2fs_destroy_compress_mempool(void) { } 4294 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4295 bool in_task) { } 4296 static inline void f2fs_end_read_compressed_page(struct page *page, 4297 bool failed, block_t blkaddr, bool in_task) 4298 { 4299 WARN_ON_ONCE(1); 4300 } 4301 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4302 { 4303 WARN_ON_ONCE(1); 4304 } 4305 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4306 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4307 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4308 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4309 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4310 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4311 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4312 static inline void f2fs_destroy_compress_cache(void) { } 4313 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4314 block_t blkaddr) { } 4315 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4316 struct page *page, nid_t ino, block_t blkaddr) { } 4317 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4318 struct page *page, block_t blkaddr) { return false; } 4319 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4320 nid_t ino) { } 4321 #define inc_compr_inode_stat(inode) do { } while (0) 4322 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4323 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4324 unsigned int c_len) { } 4325 #endif 4326 4327 static inline int set_compress_context(struct inode *inode) 4328 { 4329 #ifdef CONFIG_F2FS_FS_COMPRESSION 4330 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4331 4332 F2FS_I(inode)->i_compress_algorithm = 4333 F2FS_OPTION(sbi).compress_algorithm; 4334 F2FS_I(inode)->i_log_cluster_size = 4335 F2FS_OPTION(sbi).compress_log_size; 4336 F2FS_I(inode)->i_compress_flag = 4337 F2FS_OPTION(sbi).compress_chksum ? 4338 1 << COMPRESS_CHKSUM : 0; 4339 F2FS_I(inode)->i_cluster_size = 4340 1 << F2FS_I(inode)->i_log_cluster_size; 4341 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4342 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4343 F2FS_OPTION(sbi).compress_level) 4344 F2FS_I(inode)->i_compress_flag |= 4345 F2FS_OPTION(sbi).compress_level << 4346 COMPRESS_LEVEL_OFFSET; 4347 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4348 set_inode_flag(inode, FI_COMPRESSED_FILE); 4349 stat_inc_compr_inode(inode); 4350 inc_compr_inode_stat(inode); 4351 f2fs_mark_inode_dirty_sync(inode, true); 4352 return 0; 4353 #else 4354 return -EOPNOTSUPP; 4355 #endif 4356 } 4357 4358 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4359 { 4360 struct f2fs_inode_info *fi = F2FS_I(inode); 4361 4362 if (!f2fs_compressed_file(inode)) 4363 return true; 4364 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4365 return false; 4366 4367 fi->i_flags &= ~F2FS_COMPR_FL; 4368 stat_dec_compr_inode(inode); 4369 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4370 f2fs_mark_inode_dirty_sync(inode, true); 4371 return true; 4372 } 4373 4374 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4375 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4376 { \ 4377 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4378 } 4379 4380 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4381 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4382 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4383 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4384 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4385 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4386 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4387 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4388 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4389 F2FS_FEATURE_FUNCS(verity, VERITY); 4390 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4391 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4392 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4393 F2FS_FEATURE_FUNCS(readonly, RO); 4394 4395 static inline bool f2fs_may_extent_tree(struct inode *inode) 4396 { 4397 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4398 4399 if (!test_opt(sbi, EXTENT_CACHE) || 4400 is_inode_flag_set(inode, FI_NO_EXTENT) || 4401 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4402 !f2fs_sb_has_readonly(sbi))) 4403 return false; 4404 4405 /* 4406 * for recovered files during mount do not create extents 4407 * if shrinker is not registered. 4408 */ 4409 if (list_empty(&sbi->s_list)) 4410 return false; 4411 4412 return S_ISREG(inode->i_mode); 4413 } 4414 4415 #ifdef CONFIG_BLK_DEV_ZONED 4416 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4417 block_t blkaddr) 4418 { 4419 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4420 4421 return test_bit(zno, FDEV(devi).blkz_seq); 4422 } 4423 #endif 4424 4425 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4426 { 4427 return f2fs_sb_has_blkzoned(sbi); 4428 } 4429 4430 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4431 { 4432 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4433 } 4434 4435 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4436 { 4437 int i; 4438 4439 if (!f2fs_is_multi_device(sbi)) 4440 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4441 4442 for (i = 0; i < sbi->s_ndevs; i++) 4443 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4444 return true; 4445 return false; 4446 } 4447 4448 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4449 { 4450 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4451 f2fs_hw_should_discard(sbi); 4452 } 4453 4454 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4455 { 4456 int i; 4457 4458 if (!f2fs_is_multi_device(sbi)) 4459 return bdev_read_only(sbi->sb->s_bdev); 4460 4461 for (i = 0; i < sbi->s_ndevs; i++) 4462 if (bdev_read_only(FDEV(i).bdev)) 4463 return true; 4464 return false; 4465 } 4466 4467 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4468 { 4469 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4470 } 4471 4472 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4473 { 4474 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4475 } 4476 4477 static inline bool f2fs_may_compress(struct inode *inode) 4478 { 4479 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4480 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4481 return false; 4482 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4483 } 4484 4485 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4486 u64 blocks, bool add) 4487 { 4488 struct f2fs_inode_info *fi = F2FS_I(inode); 4489 int diff = fi->i_cluster_size - blocks; 4490 4491 /* don't update i_compr_blocks if saved blocks were released */ 4492 if (!add && !atomic_read(&fi->i_compr_blocks)) 4493 return; 4494 4495 if (add) { 4496 atomic_add(diff, &fi->i_compr_blocks); 4497 stat_add_compr_blocks(inode, diff); 4498 } else { 4499 atomic_sub(diff, &fi->i_compr_blocks); 4500 stat_sub_compr_blocks(inode, diff); 4501 } 4502 f2fs_mark_inode_dirty_sync(inode, true); 4503 } 4504 4505 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4506 int flag) 4507 { 4508 if (!f2fs_is_multi_device(sbi)) 4509 return false; 4510 if (flag != F2FS_GET_BLOCK_DIO) 4511 return false; 4512 return sbi->aligned_blksize; 4513 } 4514 4515 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4516 { 4517 return fsverity_active(inode) && 4518 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4519 } 4520 4521 #ifdef CONFIG_F2FS_FAULT_INJECTION 4522 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4523 unsigned int type); 4524 #else 4525 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4526 #endif 4527 4528 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4529 { 4530 #ifdef CONFIG_QUOTA 4531 if (f2fs_sb_has_quota_ino(sbi)) 4532 return true; 4533 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4534 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4535 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4536 return true; 4537 #endif 4538 return false; 4539 } 4540 4541 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4542 { 4543 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4544 } 4545 4546 static inline void f2fs_io_schedule_timeout(long timeout) 4547 { 4548 set_current_state(TASK_UNINTERRUPTIBLE); 4549 io_schedule_timeout(timeout); 4550 } 4551 4552 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4553 enum page_type type) 4554 { 4555 if (unlikely(f2fs_cp_error(sbi))) 4556 return; 4557 4558 if (ofs == sbi->page_eio_ofs[type]) { 4559 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4560 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4561 } else { 4562 sbi->page_eio_ofs[type] = ofs; 4563 sbi->page_eio_cnt[type] = 0; 4564 } 4565 } 4566 4567 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4568 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4569 4570 #endif /* _LINUX_F2FS_H */ 4571